Berthelot SF 98 99 175 181 PDF

Part I
Constituents and Architecture of Composite Materials
The objective of this part is to emphasize the context in which the problem of the mechanical analysis of laminate or sandwich structures is stated. Chapter 1 is an introduction which gives general features on composite materials. The constituents (matrix and fibres) are analysed in Chapter 2. The general architecture of laminate and sandwich materials is next considered in Chapter 3.
CHAPTER 1
Basic Elements on Composite Materials
1.1 COMPOSITE MATERIALS 1.1.1 Definition

In a general sense, the word composite means constituted of two or more different parts. In practice, the term composite material or composite is used in a more restrictive sense, as a material constituted by the assemblage of two or more materials of different natures with complementary properties leading to a material which have better properties than the properties of the composite components considered separately. Examples of composite materials, in this general concept, are reported in Table 1.1. The concept of composite material is specified hereafter in this chapter.
1.1.2 General Features

In the most general case, a composite material is constituted of one or more discontinuous phase distributed in a continuous phase. In the case of several discontinuous phases of different natures, the composite material is called a hybrid composite. The discontinuous phase is usually harder with mechanical properties which are much higher than those of the continuous phase. The continuous phase is called the matrix. The discontinuous phase is called the reinforcing material or the reinforcement (Figure 1.1). An exception to this composite description is the case of polymer materials modified by elastomers, for which a rigid polymer matrix is filled with elastomer particulates. For this type of composite, the static mechanical properties of the initial polymer, (Youngs modulus, strength, etc.) are not modified notably, when the impact properties are improved appreciably. The properties of composite materials result from: the properties of constituent materials, their geometrical distribution, their interactions, etc.
Chapter 1 Basic Elements on Composite Materials
TABLE 1.1. Examples of composite materials, in a general sense.

Composite Type 1. Organic Matrix Composites Paper, cardboard Particle panels Fibre Panels Coated canvas Impervious materials Tires Laminates Reinforced plastics 2. Mineral Matrix Composites Concrete Carbon-carbon composite Ceramic Composite 3. Metallic Matrix Composites 4. Sandwiches Resin/fillers/cellulose fibres Resin/wood shavings Resin/wood fibres Flexible resins/cloths Elastomers/bitumen/textiles Rubber/canvas/steel Resin/fillers/glass fibres, carbon fibres, etc. Resins/microspheres Printing, Packaging, etc. Woodwork Building Sports, building Roofing, terrace, etc. Automobiles Multiple areas Constituent Materials Application Fields
Cement/sand/granulates Carbon/carbon fibres Ceramic/ceramic fibres
Civil engineering Aviation, space, sports, biomedecine, etc. Thermomechanical applications Space
Aluminium/boron fibres Aluminium/carbon
Peaux Ames
Metals, laminates, etc. Foams, honeycombs, balsa, reinforced plastics, etc. Multiple areas
matrix reinforcement
FIGURE 1.1. Composite Material.
1.2 Classification of Composite Materials
Thus, for the characterisation of a composite material, it will be necessary to consider: the nature of the constituent materials and their properties, the reinforcement geometry, its repartition, the nature of the interface between the reinforcement and the matrix. The reinforcement geometry will be described by the reinforcement shape, its size, the reinforcement concentration, its orientation, etc. If all of these parameters combine to determine the resultant properties of the composite material, the decriptive models will consider only some of these parameters because of the complexity of the mechanical phenomena which are involved. For example, the shape of the reinforcement will be approximated as either spheres or cylinders. The concentration of the reinforcement is usually measured by the volume fraction or the weight fraction. The development of the models considered in this book will show that the volume fraction of reinforcement is a determining parameter for the mechanical properties of composite materials. For a given reinforcement concentration, the reinforcement distribution through the volume of composite materials is also an important feature. A uniform distribution will lead to the homogeneity of materials: the mechanical properties of materials will be independent of the measurement point. In the case of nonuniform distributions of the reinforcements, material fracture will be initiated in the areas of low reinforcement concentrations, which leads to a decrease in the fracture properties of composites. For composite materials in which the reinforcement is constituted of fibres, the fibre orientation is a determinant factor for the anisotropy of materials. This aspect constitutes one of the fundamental properties of composite materials: the ability to tailor the composite structure according to a conception and fabrication of the structure in such a way to obtain the structure properties wished.
1.2 CLASSIFICATION OF COMPOSITE MATERIALS

Composite materials can be classified according to the form of constituent materials or according to the nature of constituent materials.
1.2.1 Classification according to the Form of Constituents

As a function of the form of the constituents, composite materials are classified into two general classes: composite materials with fibres and composite materials with particles.
1.2.1.1 Fibre Composites

A composite material is a fibre composite when the reinforcement is in the form of fibres. The fibre reinforcement can be either continuous or discontinuous in form, as cloth reinforcement, chopped fibres, short fibres, etc. The
arrangement of fibres and their orientations allow us to modify the mechanical properties of composite materials, in such a way to obtain materials ranging from strongly anisotropic materials to transverse isotropic materials. Designers thus have in composite material a material the properties of which can be modified by adjusting: the nature of the constituents, the proportion of the constituents, the orientation of the reinforcement, according to the performances required. Fibre composite materials lead to high mechanical properties and justify to develop an extensive study of their mechanical behaviours. So, this type of composite materials will be considered essentially in the present textbook.
1.2.1.2 Particle Composites

A composite material is a particle composite when the reinforcement is in the form of particles. In contrast to fibre reinforcement, particles do not have a privileged dimension. Particles are generally introduced to improve particular properties of materials or matrices, such as rigidity, thermal behaviour, resistance to abrasion, decrease of shrinkage, etc. In numerous cases, particles are simply used as fillers to reduce the cost of the initial materials without degrading the initial properties. The choice of the particle-matrix association depends on the properties wanted. For example, lead particles in copper alloys make them easier to machine. Particles of brittle metals such as tungsten, chromium and molybdenum, incorporated in ductile materials, improve the material properties at high temperatures while preserving their ductility at room temperatures. Cermets are also examples of particular composites constituted of ceramic particles incorporated in a metal matrix. These composites are adapted to high temperature applications. For example, oxide-based cermets are used for highspeed cutting tools and for the protectors at high temperatures. Also, elastomer particles can be incorporated in brittle polymer matrices in such a way to improve their fracture and shock properties by decreasing the sensibility to cracking initiation and development. Particle composites cover an extensive domain that is constantly expending. In view of the diversity of these materials, particle composites will not be studied in the present textbook.
1.2.2 Classification according to the Nature of Constituents

According to the nature of the matrix, composite materials are classified as organic, metallic or mineral matrix composites. Various reinforcements can be associated with these matrices. Only some associations have an actual industrial interest. Other materials are subjects of developments in research laboratories.
1.3 Why Composite Materials?
Among these various composites, we can cite: 1. Composites with organic matrix (resin, fillers), with: mineral fibres: glass, carbon, etc. organic fibres: Kevlar, polyamides, etc. metallic fibres: boron, aluminium, etc. 2. Composites with metallic matrix (light and ultra-light alloys of aluminium, magnesium, titanium), with : mineral fibres: carbon, silicon carbide (SiC), metallic fibres: boron, metallo-mineral fibres: boron fibres coated with silicon carbide (BorSiC). 3. Composites with mineral matrix (ceramic), with: metallic fibres: boron, metallic particles: cermets, mineral particles: carbides, nitrides, etc. Composite materials with an organic matrix can be used only for temperatures which do not exceed 200 to 300 C. For higher temperatures, composite materials with a metallic matrix are used up to 600 C and composites with a metallic matrix up to 1,000 C.
1.3 WHY COMPOSITE MATERIALS ?

We have reported the ability of composite materials to be tailored as a function of the applications. Other reasons justify their development. We give some elements in this section.
1.3.1 Specific Mechanical Characteristics

We consider a beam loaded with a tensile load F (Figure 1.2). The relation between the load and the elongation l of the beam is given by:
F=
ES l , l
(1.1)
where E is the Youngs modulus of the beam material, S the cross-sectional area of the beam and l the length of the beam. The beam stiffness K = ES l caracterizes the mechanical performances of the beam in the elastic domain. In the case of two materials 1 and 2, the ratio of the
FIGURE 1.2. Tensile loading of a beam.
beam stiffnesses is: K1 E1S1 l2 = , K 2 E2 S2 l1 and the ratio of the weights of the beams is: m1 S1l1 1 = , m2 S2l2 2 (1.3) (1.2)
introducing the specific weights (weights per unit of volume) of the beams. The combination of Relations (1.2) and (1.3) leads to : K1 E1 1 m1 l2 = . K 2 E2 2 m2 l1
2
(1.4)
For a structure, the dimensions of the elements are given, and the comparison of the beam stiffnesses must be considered for identical lengths. Therefore, for l1 = l2 : K1 E1 1 m1 = . (1.5) K 2 E2 2 m2
Lastly, the use of materials in the space and aviation areas, and further in the areas of sports, building, etc., has led to compare the mechanical properties of structures with equal weights. For m1 = m2 , the stiffness ratio is: K1 E1 1 = . K 2 E2 2 (1.6)
Thus, it appears that the best material is that which has the highest value of E , leading to the highest value of the stiffness of the beam. The term E is called the specific Youngs modulus of the material. A similar investigation can be implemented in the case of a three-point bending beam where the beam is subjected to a load F (Figure 1.3). The relation between
1.3 Why Composite Materials?
the load and the transverse deflection at the beam centre is given by : (1.7) f = Kf , l3 where f is the deflection at the beam centre, I the inertia moment of the cross section and l the span length distance between the supports. The coefficient K is the bending stiffness of the beam. In the case of a cylindrical beam section of radius r, the moment is given by
and the weight is I=
F = 48 EI
r4,
m = r 2l.
It follows that in the case of two materials 1 and 2, the bending stiffness ratio is given by:
K1 E1 12 m1 l2 = . 2 K 2 E2 2 m2 l1
2 5
(1.8)
So, for bending beam, it results that the best material is that which has the highest value of E 2 . Similar developments can be considered in different shapes of the structures: plates, shells, complex structures. The conclusion is always of the same nature: for identical weights and dimensions, the most rigid structures are obtained by using materials that have the smallest specific weight. Similarly, the comparison of the structure strengths lead to similar conclusions for the fracture stresses of materials. Thus, it has become usual to compare the mechanical properties of materials by considering the specific values (with respect to the weight per unit of volume) of the moduli and fracture stresses of materials.
1.3.2 Mechanical Characteristics of Materials

From the previous considerations, we now look for the most efficient materials: high modulus and fracture stress, low density. It is also obvious that the elaboration of these materials must not result in a high cost, which itself depends on their area of use. For example, in space and aviation applications, high properties are sought and the cost of materials and their elaboration has a low impact. In contrast, in the automotive industry, the improvement of the F
l
FIGURE 1.3. Three-point bending beam.
10
performances can not be achieved to the detriment of the cost of the finished product. In this area, the impact of the cost of the material and its elaboration is very high. Table 1.2 gives the specific mechanical characteristics of usual materials elaborated in bulk form. The traditional materials such as steel, aluminium alloys, wood and glass have comparable specific moduli. In contrast, it is observed that the specific fracture stress of glass is clearly higher than that of steel and of aluminium alloys. Furthermore, it is an established fact that the fracture stresses measured for the materials are notably smaller that the theoretical values. This difference is attributed to the presence of defects, as microcracks, in the materials. To increase the values of the fracture stresses, it is then necessary to seek for processes of material elaboration which lead to a decrease of the defects inside the materials. This objective is achieved by elaborating the materials in the form of fibres of very small diameters of some tens of microns. It is clear that it is necessary to proceed from materials which already have high specific properties in the bulk form. The mechanical properties of materials elaborated in the form of fibres are reported in Table 1.3. The values reported clearly show the interest in elaborating materials in fibre form to achieve higher values of the fracture stresses. Owing to their low cost, glass fibres are used most where the low cost of the products is a determinant factor. However, glass fibres have a limited value of the modulus. Other fibres, as carbon fibres, Kevlar fibres and boron fibres have a high specific modulus, hence the interest of these fibres in space and aviation areas.
1.3.3 Composite Materials

Because of their low cross sections (diameters of 10 to 20 m), fibres cannot, however, be used directly in mechanical applications. Whence the idea of incorporating them in a polymer matrix in order to make a fibre composite material. The matrix then has various functions: to link the fibres together, to transfer the mechanical loads to the fibres, to protect the fibres from the external environment, etc. Thus, a new material is born that is adjustable and that has high specific mechanical characteristics. The components and the general structure of composite materials will be studied in more detail in Chapter 2.
1.4 VOLUME AND WEIGHT FRACTIONS 1.4.1 Introduction

One of the most important factors which determine the mechanical properties of a composite material is the relative proportions of reinforcement and matrix. The constituent proportions can be evaluated either by the volume fractions or by the weight fractions. The weight fractions are easier to determine when the
1.4 Volume and Weight Fraction
11
TABLE 1.2. Specific properties of usual materials in bulk form.

Modulus Material E (GPa) 210 70 30 70 350 300 Fracture Stress Density (kg/m3) 7,800 2,700 390 2,500 19,300 1,830 Specific Modulus E / (MN m/kg) 26.9 25.9 33.3 28 18.1 164 Specific Stress
(MPa) 3402,100 140620 700-2,100 1,1004,100 700
(kN m/kg) 43270 52230 280840 57210 380
Steel Aluminium alloys Wood Glass Tungsten Beryllium
TABLE 1.3. Specific mechanical properties of materials, elaborated in the form of fibres.
Modulus Fibres E (GPa) 72.4 85.5 390 240 130 385 Fracture Stress Density (kg/m3) 2,540 2,480 1,900 1,850 1,500 2,630 Specific Modulus E / (MN m/kg) 28.5 34.5 205 130 87 146 Specific Stress
(MPa) 3,500 4,600 2,100 3,500 2,800 2,800
(kN m/kg) 1,380 1,850 1,100 1,890 1,870 1,100
E-Glass S-Glass Carbon with high modulus high stress Kevlar (aramid) Boron
composite materials are elaborated. Moreover, the analysis of the mechanical properties which will be developed in this book will show that the volume fractions are the factors which it is necessary to introduce in the theoretical models for describing the mechanical properties of composites. It is therefore necessary to derive the relations which relate one fraction to the other. These relations will be established for a two-phase material and then extended to a material with multiple phases.
1.4.2 Volume Fractions

Let us consider a volume vc of composite material, constituted of a volume vf of fibres and a volume vm of matrix. Indices c, f and m will be used as the
12
respective indices for the composite material, the fibres and the matrix. The volume fraction of fibres is: v Vf = f . (1.9) vc The volume fraction of the matrix is:
Vm =
with
vm , vc
(1.10) (1.11)
Vm = 1 Vf ,
since
vc = vf + v m .
(1.12)
1.4.3 Weight Fractions

The weight fractions are defined in a similar way introducing the respective weights pc, pf, pm of the composite material, the fibres, the matrix. The weight fractions of fibres and matrix are given respectively by:
Pf = Pm = pf , pc pm , pc
(1.13) (1.14) (1.15)
with
Pm = 1 Pf .
1.4.4 Relations between Volume and Weight fractions

The relations between the volume fractions and weight fractions introduce the respective specific weights c, f, m of composite, fibres, matrix. The weights and volumes are related by expressions:
pc = cvc , pf = f vf , pm = mv m .
(1.16) (1.17) (1.18)
The total weight of composite material is:

pc = pf + pm ,
or
cvc = f vf + mv m .
The specific weight of the composite material is thus written as a function of the volume fractions as:
1.4 Volume and Weight Fraction
13
c = f Vf + m (1 Vf ) .
Similarly, considering the total volume of the composite:
vc = vf + v m ,
(1.19)
(1.20) . (1.21)
we obtain:
pc
pf
pm
Whence the expression for the specific weight of the composite:
c =
Pf
Pm
(1.22)
The relations between the weight fractions and volume fractions can now be established considering the relations:
Pf = pf f vf f = = Vf , pc cvc c Pm =
(1.23)
and
m Vm , c
(1.24)
where the specific weight of composite material is deduced from Relation (1.19). The inverse relations are obtained in a similar way. We have:
Vf = Vm =
c Pf , f c Pm , m
(1.25) (1.26)
where the specific weight of composite material is deduced from Relation (1.22). Equations (1.19) to (1.26) can be extended to the case of an arbitrary number of constituents. The genaral expressions for n constituents are:
Pi =
i Vi , c
n
(1.27)
with
c =
and
iVi ,
i =1
(1.28)
Vi =
c Pi , i
(1.29)
with
14
c =
i =1 i
n
Pi
(1.30)
1.4.5 Presence of Porosity

In practice, the specific weight (density) measured experimentally does not coincide exactly with the values derived from Expression (1.22) when the constituent weights are introduced. In the case where the difference exceeds the experimental errors, this difference can be attributed to the presence of porosity in the composite material. The difference between the density ct deduced from xpression (1.22) and the density ce measured experimentally allows us to evaluate the volume fraction Vp of the porosities by the Expression: ce Vp = ct . (1.31)
ct
The presence of porosities in a composite may involve a significant decrease of the mechanical properties of the composite. Porosity also increases the sensitivity of the composite material to the external environment: increase of the humidity absorption, decrease of the resistance to chemical products, etc. So, it is important to have an evaluation of the porosity proportion as a means to estimate the quality of a composite. A high-quality composite material will contain less than 1 % by volume of porosities, when a mediocre-quality composite could be reached as much as 5 %.
EXERCISES
1.1 Express the volume fraction Vf of fibres in a composite as a function of the weight fraction, introducing the ratio f /m of the specific weights and the ratio (1 Pf) / Pf of the weight fractions of matrix and fibres. 1.2 Plot the curve for the volume fraction of fibres as a function of the weight fraction of fibres in the case of glass fibre (f = 2500 kg/m3) composites, of carbon fibres (f = 1900 kg/m3), of Kevlar fibres (f = 1500 kg/m3), for the same matrix m = 1200 kg/m3. 1.3 A composite structure is designed as made of a composite containing a volume fraction Vf of fibres. The volume of the structure is vc. Calculate the fibre and matrix weights which are necessary. Application : Vf = 50%, vc = 0,01 m3. Calculate the weights in the case of the composite materials considered in Exercise 1.2.
CHAPTER 2
The Constituent Elements of Composite Materials
2.1 INTRODUCTION
As considered in this book, a composite material is constituted of a matrix with fibre reinforcement embedded in the matrix The matrix is a resin, such as polyester, epoxide, etc., in which fillers are incorporated, in such a way to reduce the production cost while improving the properties of the resin. From a mechanical point of view, the filler-resin system behaves as a homogeneous material. So, the composite material is considered as being constituted of a matrix and a fibre reinforcement. The reinforcement brings to the composite material its high mechanical properties, when the role of the matrix is to transfer to the reinforcement the external mechanical loading and to protect the reinforcement against external aggressions. The type of reinforcement-matrix association depends on the constraints imposed by the designer: high mechanical properties of composite, good thermal stability, low cost, resistance to corrosion, etc. The purpose of this chapter is to give a general view of the various constituents which are used. The synthesis is given from the viewpoint of the mechanical engineering. For a more extensive analysis, the reader should refer to specialized books.
2.2 THE RESINS 2.2.1 Types of Resins

The resins used in composite materials play the role of transferring the external mechanical loading to the reinforcement and to protect it from the external environment. The resins must therefore be quite flexible and offer a good compatibility with the reinforcement. In addition, they must have a low density to keep the high specific properties of the reinforcement.
16
Chapter 2 The Constituent Elements of Composite Materials
According to these considerations, the resins used for composite materials are polymers modified by fillers and additives, such as mould release agents, stabilizers, pigments, etc. Resins are delivered in solution in the form of polymers in suspension in solvents that prevent linking between the prepolymerized macromolecule. When heated, links are developed between the chains of the prepolymer so as to constitute a cross-linked polymer with a three-dimensional structure. Two large families of polymer resins exist: thermoplastic resins and thermosetting resins. These two types of resin have the property of being able to be moulded or manufactured in order to obtain either a finished product or a semifinished product the form of which can be next modified. The thermoplastic resins, the production of which reaches the highest tonnage because of their low cost, have the property of being able to be processed several times by successive heating and recooling. These resins can thus be salvaged and easily recycled. However, the mechanical and thermal properties are low. In contrast, the thermosetting resins can be processed only once. After polymerization by heat applying in the presence of a catalyst, these resins lead to a structure that can be destroyed only by an application of high thermal energy. Thus, thermosetting resins have mechanical properties and especially thermomechanical properties which are much higher that those of thermoplastic resins. As a result of these higher properties, thermosetting resins are extensively used in the manufacture of composite materials. However, improvement of the properties of thermoplastic resins leads to their increasing use. Two other classes of resins with specific uses are also used in the elaboration of composite materials. They are: thermostable resins which in continuous service can support temperatures of the order of 200 C and higher, elastomers, the reinforcement of which using different fibres leads to various applications in the automotive industry.
2.2.2 Thermosetting Resins

The principal thermosetting resins used in manufacturing composite materials are: unsatured polyester resins: condensed polyesters, vinylesters, allylic derivatives, etc., condensation resins: phenolics, aminoplasts, furanes, etc., epoxide resins.
2.2.2.1 Polyesters Resins

The most widely used of all the resins in the manufacture of composite materials are the unsatured polyester resins. Their development is the result of the following characteristics:
2.2 The Resins
17
low production cost, a wide diversity that offers many possibilities, their adaptation to different fabrication processes that are easy to carry out and to automatize. Hence the industrial development of polyester resins is continually increasing. According to their Youngs modulus, polyesters resins are classified into flexible resins, semirigid resins or rigid resins. The resins usually used in manufacturing composite materials are of rigid type. These cured polyester resins have the following properties: Density Tensile modulus Bending modulus Tensile fracture stress Bending fracture stress Tensile fracture strain Bending fracture strain Compressive fracture stress Shear fracture stress Deflection temperature under a load 1,200 kg/m3 2.83.5 GPa 34.5 GPa 5080 MPa 90130 MPa 25 % 79 % 90200 MPa 1020 MPa 60100 C
Among the advantages of the unsatured polyester resins are the following : a good rigidity, resulting from a fairly high Youngs modulus, a good dimensional stability, a good wettability of reinforcements, the ability to be manufactured, a good chemical behaviour, a low cost of production, a good chemical resistance to hydrocarbons (petrol, fuel, etc.) at room temperatures, etc. Among the disadvantages, it can be reported: a mediocre behaviour with temperature, which is less than 120 C in continuous use, a sensitivity to cracking, especially under shocks, an inportant shrinkage, of the order of 8 to 10 %, a poor behaviour in steam environment, with the risk of hydrolysis, hence the necessity of covering polyester resin composites with a gel-coat layer to protect them, a degradation in ultraviolet environment, a flammability.
2.2.2.2 Condensation Resins

Condensation resins comprise three types of resins: the phenolic resins, the aminoplasts and the furane resins.
18
1. Phenolic resins are the oldest of the thermosetting resins, the best known of which is the bakelite. The characteristics of these resins are the following ones: Density Bending modulus Tensile fracture stress Tensile fracture strain Bending fracture stress Compressive strength Deflection temperature under a load Among the advantages, we have: an excellent dimensional stability, a good temperature stability, a good chemical resistance, a low shrinkage, good mechanical properties, a low cost. Among the disadvantages, we report: an elaboration using moulding processes with pressure, hence low production rate, dark colors of resins, the resins can not be used for food applications. Phenolic resins will therefore used for applications that require a high temperature behaviour or good chemical resistance. 2. The properties of aminoplast resins are close to those of phenolic resins. To the advantages of these resins, it can be added: the ability to use in food applications, the possibilty of colouring the resins. 3. Furane resins provide probably the best chemical resistance of any thermosetting resin. Hardening is faster than for phenolic resins. However the cost of furane resins is high, about three times higher than the cost of phenolic resins, hence the use of the furane resins is limited. Their good chemical resistance leads to use the furane resins for tanks, pipes, containers of chemical agents. 1,200 kg/m3 3 GPa 40 MPa 2.5 % 90 MPa 250 MPa 120 C
2.2.2.3 Epoxide Resins

The epoxide resins are used most widely after the unsaturated polyester resins. However they represent only of the order of 5 % of the composite market on account of their high price, of the order of five times more than polyester resins. Because of their good mechanical properties, the epoxide resins, usually used without fillers, are the matrices of composite materials with high mechanical
2.2 The Resins
19
performances which are used in the areas of aeronautical construction, space, missiles, etc. The general mechanical characteristics of epoxide resins are the following ones: Density Tensile modulus Tensile fracture stress Bending fracture stress Tensile fracture strain Shear fracture stress Deflection temperature under a load 1,1001,500 kg/m3 35 GPa 6080 MPa 100150 MPa 25 % 3050 MPa 290 C
Epoxide resins thus lead to high mechanical properties. However, to benefit of these high performances, it is necessary to have long cycles of transformation and long cure times, from several hours to several tens of hours, at relatively high temperatures from about 50 to 100 C. Among the advantages of epoxide resins, we have: good mechanical properties in tension, compression, bending, shock, etc. superior to those of polyester resins, a good behaviour at high temperatures: up to 150 C190 C in continuous use; an excellent chemical resistance, a low shrinkage in moulding processes and during cure (about 0,51 %), a very good wettability of reinforcements, an excellent adhesion to metallic and mineral materials. Among the disadvantages, it can be reported: a long time of polymerization, a high cost, the need to take precautions in the manufacture processes, a sensitivity to cracking.
2.2.3 Thermoplastic Resins

The family of thermoplastic resins, currently called as plastics, is quite wide and can be separated into plastics used widely in current applications and technical plastics or technopolymers. The plastics of current applications are processed either by injection to obtain moulded objects, or by extrusion to obtain films, plates, tubes, profiled objects, etc. The technical plastics are usually processed by injection. Among thermoplastics are: the polyvinyl chloride (PVC), the polyethylene,
20
the polypropylene, the polystyrene, the polyamide, the polycarbonate, etc. The interest of thermoplastics lies in their low cost, resulting from the initial materials as well as the fabrication processes (injection, extrusion, pultrusion, etc.). Nevertheless, this low cost is associated with mechanical and thermo-mechanical properties which are low. We report hereafter some characteristics for Polypropylene and polyamide. Polypropylene Density (kg/m3) Fracture stress (MPa) Tensile modulus (GPa) Deflection temperature under a load (C) 900 2035 1.11.4 5060 Polyamide 1,140 6085 1.22.5 65100
The various thermoplastics can be reinforced by fibres to obtain composite materials. However, the use of thermoplastics in the domain of composites have a limited development on account of the low properties of thermoplastic resins and the need to use high temperatures to process the transformations from initial solid products.
2.2.4 Thermostable Resins

Thermostable resins differ from the other resins previously considered, essentially by their thermal performances which keep the mechanical properties of composite structure for temperatures higher than 200 C. In practice, we find in these resins the two families of thermoplastic and thermosetting resins. The thermostable resins have been developed in the aviation and space areas, where the research laboratories are continuously developing new resins. Among the thermostable resins, bismaleide resins and polyimide resins are the most used. Bismaleide resins are resins whose the cross-linked structure is elaborated at temperatures between 180 C and 200 C. The moulding processes are those used for composites with thermosetting resins. Polyimide resins appear on the market around 1970. The cross-linked structure of the resins is obtained at high temperatures, between 250 C and 300 C, and the elaboration needs a polycondensation which gives reaction products. So, these resins allow us to obtain composites with mechanical properties at 250 C which are higher than those of aluminium alloys. But the manufacturing processes are complicated leading to high cost materials.
2.3 FILLERS AND ADDITIVES 2.3.1 Introduction

Different products can be incorporated into resins to modify the initial resin
2.3 Fillers and Additives
21
systems or to reduce the final costs of the composite matrices. The quantity of the added products can vary from some tens of percent in the case of fillers to a few percent or less in the case of additives. The addition of these products has the function of improving the mechanical and physical properties of the finished products as well as making easier their manufacture processes. Some example of fillers and additives are given hereafter in this section.
2.3.2 Fillers
2.3.2.1 Reinforcing Fillers
The object of incorporating reinforcing fillers in resins is to improve the initial mechanical properties of resins. These fillers can be classified according to their geometric forms, as spherical fillers or non-spherical fillers. 2.3.2.1.1. Spherical Fillers The basic interest in these fillers lies in their spherical forms which avoid stress concentrations and, as a consequence, decrease the cracking susceptibility of the matrix compared with the use of non-spherical fillers. Spherical fillers, usually called microspheres, are elaborated in the forms of solid spheres or hollow spheres. The solid or hollow microspheres have diameters usually ranging from 10 to 150 m. They can be elaborated from glass, carbon or organic component as epoxide, phenolic, polystyrene, etc. Hollow glass microspheres represent more than 90 % of the spherical fillers used for composite matrices. Hollow Glass Microspheres The main advantage of hollow glass microspheres lies in their low density, from 100 to 400 kg/m3, yielding an increase in the specific modulus of the filled resins and their behaviour under compression. The fabrication of hollow glass microspheres is carried out by flowing, in a high temperature area, fine particles of glass containing an expansion gas, usually a mixture of nitrogene and carbon dioxide. When the particles are submitted to the high temperatures, the gas expands inside the melted glass particles. Next, the particles are rapidly cooled inducing the solidification of the walls of the microspheres before the decrease of the gas pressure. The hollow spheres so obtained has diameters of the order of 20 to 130 m, with wall thicknesses of about 0.5 to 2 m. The microspheres can next be selected according to their sizes. Lastly, the spheres are submitted to surface treatments, leading to the improvement of bonding between resin and spheres. Hollow glass microspheres are most frequently incorporated in epoxide or polyester resins. Their use is restricted to low pressure manufacturing processes because of the weak resistance of the hollow spheres to crushing.
22
The essential advantages of incorporating hollow glass spheres in resins are: the decrease of the matrix density, an increase of the Youngs modulus of the resin, the improvement in the behaviour of matrix under compression. Other microspheres Other hollow microspheres are elaborated as: Carbon microspheres: density 120 kg/m3, diameter 5 to150 m. Organic microspheres (epoxide, phnolic, etc.) : density 100 to 500 kg/m3, diameter 10 to 800 m. These microspheres are usally more expensive (up to five times more for carbon microspheres) than glass spheres. Among other microspheres used are solid glass spheres. Compared to hollow glass spheres, the characteristics of solid glass spheres are: a high density: 2,500 kg/m3, a lower cost, the ability of manufacturing the products at high pressures. 2.3.2.1.2. Non-spherical Fillers Among the non-spherical fillers, mica is the material which is the most used. It is then incorporated in the forms of flakes of transverse dimensions from 100 to 500 m and thicknesses of 1 to 20 m. Mica is usually added to thermoplastic or thermosetting resins for electrical and electronical applications.
2.3.2.2 Non-Reinforcing Fillers

Non-reinforcing fillers have the role of either reducing the cost of resins while preserving their initial performances, or of improving some properties of the resins. 2.3.2.2.1. Low Cost Fillers These fillers are extracted from rocks or minerals, hence their low cost. In practice, incorporating these fillers leads to: an increase in: the density of the matrix, the Youngs modulus, the hardness, the viscosity, the dimensional stability; a decrease in: the elaboration cost of the matrix, the tensile and bending fracture.
2.3 Fillers and Additives
23
The principal fillers which are used are: carbonates: chalks or calcites (CaCO3), which are the most used, silicates: talc, kaolin, feldspar, wollastonite, silicas, obtained by crushing and sifting of quartz sand. 2.3.2.2.2. Fire Retardant Fillers These fillers are incorporated in resins for reducing or impeding the combustion processes. Among the solid fillers introduced in thermosetting resins are: the aluminium hydrate, the most used, the antimony oxide. 2.3.2.2.3. Conductive and Anti-Static Fillers Organic resins are thermal and electrical insulators. For specific applications, it is therefore to incorporate in resins conducting fillers. The usual fillers are: metallic powders or flakes of copper, aluminium, iron, etc., glass spheres metallized with copper, silver, etc., carbon particles, as carbon black, metallic filaments.
2.3.3 Additives
Additives are incorporated in small quantities (a few percent or less) and processed as: mould release agents, pigments and dyes, anti-shrinkage agents, light stabilizers.
2.3.3.1. Mould Release Agents

Mould release agents are introduced to make easier demoulding of the composite structures from the mould, mandrels, etc. In the case of porous mould surface, a mould sealing compound has to be applied first.
2.3.3.2. Pigments and dyes

Pigments are insoluble products occuring in the forms of powders or flakes. Usually, they are derived from oxides or metallic salts. Starting from these pigments, it is also possible to obtain colorant pastes elaborated from pigments dispersed in a paste (resin or plastifier) so as to be easily used. Dyes are organic components which are soluble in a suitable solvent. Their use is generally limited due to their poor chemical and thermal behaviours.
24
2.3.3.3. Low Shrink and Low Profile Agents

Polymerization of resins leads to a decrease in the interatomic distances of the initial monomer. It results a shrinkage of the resin, which can induce a poor finish quality of the surface, warping or microcracking in the moulded pieces. Although the incorporation of fillers in the resins limit shrinkage, it is often necessary to add specific low shrink and low profile additives. These additives decrease or remove the shrinkage processes, improving also the flow of resins in moulding processes. These additives are usually elaborated from thermoplastic or elastomer elements which are in the forms of powders or in solutions in styrene.
2.3.3.4. Light Stabilizers

Light stabilizers have the function to protect resins when they are exposed to ultraviolet radiations in the case of prolonged exposure to sunlight. The principle of these agents is to absorb the radiations and to avoid premature degradation of resins through the ruptures of links between atoms.
2.4 FIBRES AND CLOTH REINFORCEMENTS 2.4.1 General Features

Reinforcements provide to composite materials their high mechanical properties: stiffness, strength, hardeness, etc. The reinforcement also allows to improve some of the physical properties, as thermal properties, fire resistance, abrasion resistance, electrical properties, etc. The characteristics which are required for reinforcements are high mechanical properties, low density, good compatibility with matrices, ability of manufacturing, low cost, etc. According to the application areas, reinforcements can have diverse origins, as vegetable, mineral, organic, synthetic, etc. However, the most widely used reinforcements are in the form of fibres or derived forms. Usually reinforcements constitute a volume fraction of composites which ranges from 0.3 to 0.7. Reinforcements are supplied in various commercial forms, as: linear forms (strands, yarns, rovings, etc.), surfacing cloths (woven fabrics, mats, etc.), multidirectional forms (preforms, complex cloths, etc.). Reinforcements can be also obtained using discontinuous fibres, as short fibres. Particular short fibres, called whiskers, with high mechanical properties, have also been developed. Their use is however limited due to the difficulties of fabrication.
2.4.2 Fibre Forms

Fibres are elaborated with a diameter of a few microns (about 10 m) and consequently they cannot be used in a single form. For their practical use, fibres
2.4 Fibres and Cloth Reinforcements
25
are gathered together into a bundle called a strand. The usual nomenclature of the various fibre forms is still not well established, and is generally derived from the nomenclature used for glass fibres. A single continuous fibre is usually called an elementary filament or monofilament. Monofilaments are next gathered into strands or yarns. Continuous or discontinuous yarns are characterised by their linear density, which is the weight per unit length. This linear density is a measure of the fineness of the strands or yarns, and depends on the diameter and number of monofilaments. The linear density is given by the tex number, which is the weight of a strand or yarn of length 1,000 m. Hence: 1 tex = 1 g/km. In fact, it would be better to write: 1 tex = 10-6 kg/m. according to SI system of units. The first definition is best adapted to a practical use.
2.4.3 Surfacing Forms

Strands and yarns can be used to make surface tissues of various types, as mats, woven fabrics or ribbons. Initially these forms were developed in the case of glass fibres.
2.4.3.1. Mats
Mats are layers of continuous or discontinuous strands or yarns randomly distributed in a plane without preferential orientation. Fibres are bounded together with either a high or low solubility binder. The absence of preferential orientation of fibres leads to mechanical properties of layers which are isotropic in the plane of the mat. The difference between chopped strand mats and continuous strand mats lies essentially at the level of their deformability properties. The first ones are not very deformable, whereas the second ones can be processed to obtain complex shapes by a regular stretching of the mat in all the directions. Continuous strand mats are particularly suitable for matched-die moulding with deep moulds to process complex shapes of structures by pressure, injection or vacuum moulding.
2.4.3.2. Woven Fabrics and Ribbons

Two-dimensional (2D) woven fabrics consist of two sets of interlaced strands or yarns (Figure 2.1) which are processed by a loom equipment. The fibre sets arranged in the lengthwise is called the warp and the fibre set in the crosswise is the weft (or fill).
26
weft
warp FIGURE 2.1. Warp and weft of a woven fabric.
Woven fabrics differ by the type of fibres (strands, yarns, rovings, etc.) and thus by the linear density of warp, as well as by the type of weave. Different types of weaves can be identified according to the repeating pattern of warp and fill interlacing. Figure 2.2 gives some examples of usual weave styles: plain weave or taffeta, twill or serge weave, satin weave, cross-ply weave, unidirectional weave. Plain Weave In plain weave, also known as taffeta, each warp and fill thread passes over one thread and under the next, leading to a cloth which is plane and stable, but not very deformable. This weave fabric leads to mechanical properties fairly identical in the two warp and weft directions, when the weaving threads are the same. However, the plain weave fabrics induce a high degree of crimp to the fibres, leading to a decrease of some mechanical performances of composite materials. Twill Weave In twill weave, known as serge, the number of warp threads and weft threads which interlace with each other can be varied. In a 2 1 serge, weft threads pass over one and under two warp threads, and in a 2 2 serge, the weft threads pass over two and under two warp threads. This type of weave fabric leads to a diagonal pattern of the weave (Figure 2.16). Serge cloth allows slippage to occur between warp and weft and has a good adaptability to moulding processes in the case of complex shapes. Satin Weave Satin weave is quite similar as serge, but the number of warp threads and weft threads that pass over each other before interlacing is greater. Each satin weave is characterised by a number, usually 4 or 8, indicating that the warp threads pass over 4 or 8 weft threads. This results in one face of the satin cloth which contains mostly of warp threads, and the other face of weft threads. Satin weave shows excellent ability for moulding complex shapes.
2.4 Fibres and Cloth Reinforcements
27
Plain weave or taffeta
2 2 twill weave or serge
8 satin weave
Cross-ply weave
Unidirectional cloth
FIGURE 2.2. Examples of fabric weaves.
Cross-ply Weave In cross-ply weave, two layers of cross threads are superposed without interlacing and held together by fine warp and weft threads. The absence of interlacing suppresses the shear effects and leads high performances of cloth, but with a high cost. Unidirectional Cloth In unidirectional cloth, the threads are aligned in the warp direction, and held together by fine weft threads. So, the fabric is unidirectional with high performances in the warp direction. The mechanical properties of the various cloths depend on: the type of threads used to manufacture the cloth: their nature (glass, carbon, etc.), the type of assemblage (with or without twisting), etc.; the type of weave: unidirectional weaves lead to the best mechanical properties of composite materials along a given direction; satin weaves and, to a lesser degree, twill weaves lead to mechanical properties which are greater than the ones of plain weaves; the linear density of the warp and weft threads of the cloth.
28
2.4.4 Multidirectional Woven Structures

2.4.4.1 Preforms
It is possible to manufacture preforms by processing a cylindrical or conical weaving to obtain a preform cloth (Figure 2.3). The threads interlace helically, and the variation in the pitch allows to adjust the preform to the shape of the structure to be moulded. It is thus possible to mould a structure having a variable diameter along a given direction. By this process, various preforms can be obtained in conical, pointed arch, hemispherical, etc., shapes, essentially used for aeronautical applications, as pipes, reentry cones, etc.
2.4.4.2 Multidirectional Weaves

Volume weavings are also develops and are characterised by the number of the weaving directions: 3D, 4D, etc. The simplest structure is that of 3D weaving in which the threads are arranged in three orthogonal directions (Figure 2.4). In a 4D weaving, the threads are arranged in 4 directions (Figure 2.5). The object of these volume weaving is to obtain composite materials which are isotropic.
2.5 DIFFERENT FIBRES 2.5.1 Glass Fibres

2.5.1.1 General Elements
Glass in bulk form is characterised by a great brittleness due to a high sensitivity to cracking. In contrast when elaborated in the form of fibres of small
FIGURE 2.3. Cylindrical and conical weavings.
2.5 Different Fibres
29
1 2
FIGURE 2.4. 3D orthogonal weaving. 4
4 FIGURE 2.4. 4D weaving.
diameters (some tens microns), the glass loses this character and then has good mechanical properties. Glass fibres are manufactured from special mineral glasses, composed of silica alumina, lime, magnesia, etc. These low cost materials, associated to quite simple production processes, give glass fibres an excellent price/performances ratio, which put glass fibres in the first rank of reinforcements actually used in composite materials. According to their compositions, different types of spinning glasses can be used to product glass fibres (Table 2.1). In practice, glasses of type E constitute almost all the production of glass fibres which are actually produced. The other types of glasses are used for specific applications as: D-glass, with high dielectric properties, used for the construction of electronic equipments for telecommunications, in particular for radoms; C-glass, resistant to chemical agents, used for the superficial processing of structures exposed to chemical environment;
30
TABLE 2.1. Different types of spinning glasses. Type E D A C R, S General characteristics General use in mechanical engineering High dielectric properties High alkali content Good chemical resistance High mechanical properties
R- and S-glasses, with high mechanical properties, used for manufacturing composite structures with high mechanical performances. Only glass fibres of types E and R are used for mechanical applications, the compositions of which are reported in Table 2.2. It can be noted the very small proportion or the absence of akaline oxide in contrast of glasses used currently. This fact leads to production processes of glass fibres with high temperatures.
2.5.1.2 Manufacturing Glass Fibres

Glass fibres are produced (Figure 2.6) by feeding the raw glass into a fiberizing element referred to as a bushing, fabricated from platinium-rhodium alloy and pierced in its base by calibrated orifices of about 2 mm in diameter. The molten glass is kept in the bushing, heated by Joule effect, at about 1,250 C. At this temperature, the glass viscosity allows a flow under gravity though the orifices in the form of fibres of some tens of millimetres. At the exit of the bushing the glass, in a plastic phase, is simultaneously drawn at high speed and cooled. According to
TABLE 2.2. Compositions of glasses of types E, D and R. Constituents Silica Alumina Lime Magnesia Boron oxide Fluorine Iron oxide Titanium oxide Sodium oxide Potassium oxide SiO2 Al2O3 CaO MgO B2O3 F Fe2O3 TiO2 Na2O3 K2O Mass proportion (%) E-glass 5354 1415.5 2024 6.59 00.7 <1 <1 D-glass 7374 0.50.6 2223 0.10.2 1.3 1.5 R-glass 60 25 9 9
31
Alumina
Magnesia Silica Lime Boron oxide
Furnace Molten glass
Bushing
Monofilament
Sizing Strand
FIGURE 2.6. Schematic diagram of glass fibre drawing process.
the cooling conditions and speed of drawing, glass filaments with different diameters can be obtained. The monofilaments are then gathered together without twisting into a bundle called a strand, which is wound into a spool or onto a tube. These basis strands are mostly used for produce glass reinforcements incorporated in composite materials. Discontinuous glass fibres can also be obtained and next assembled into a yarn called verranne. The verranne yarns are distinguished from the continuous fibres by a cotton wool appearance.
2.5.1.3 Sizing of Glass Fibres

The glass filaments issued from the drawing process cannot be used directly for a number of reasons:
32
absence of cohesion between the fibres of a strand; sensitivity of the glass to abrasion, leading to a deterioration when the fibres are handled after drawing; sensitivity to attack by humidity and water; creation of electrostatic charges resulting from frictions between fibres, etc. To overcome these defects, an operation called sizing is carried out at the exit of the bushing, which consists in depositing on the glass fibres a sizing product the composition of which depends on the further use. Considering the defects reported, the functions of sizing process are: to establish some cohesion between the filaments; to give a greater or lesser stiffness to the strands according to strand use: strands used for moulding or weaving must be flexible, strands to be cut must be stiffer; to protect the filaments against abrasion; to avoid the generation of electrostatic charges; to make easier the impregnation of the filaments by the resins: good superficial wetting of fibres and good penetration inside the fibre strands; to promote bonding between the resin matrix and the fibres upon which the mechanical properties of composite materials are depending, but also to increase the aging behaviour of composites, to decrease the sensitivity to humidity and corrosion, etc.
2.5.1.4 Mechanical Properties of Glass Fibres

It is usual to report as references the mechanical properties which are measured on monofilaments removed at the exit of the bushing. Table 2.3 gives the nominal values of the fibre characteristics. After drawing, the glass filaments are subjected to various mechanical (handling, abrasion, etc.) and chemical (humidity, corrosion, etc.) actions, which reduce their initial mechanical properties. Table 2.4 compares the values of tensile fracture stresses, measured on monofilaments and industrial strands. The values obtained show a decrease in the characteristics when the number of filaments increase. In practice the decrease results from filament handling and from the fact that it is not easy to derive the characteristics of strands because the difficulties to
TABLE 2.3. Mechanical Characteristics of E- and R-glass fibres, measured on filaments at the exit of bushing. Characteristics Density Youngs modulus Tensile fracture stress Ultimate strain Poissons ratio E-glass R-glass 2,550 86 4,400 5.2
Ef
kg/m3 GPa MPa %
2,600 73 3,400 4.4 0.22
fu fu
vf
33
TABLE 2.4. Tensile fracture stresses measured on monofilaments and industrial strands. (in MPa). Characteristics Monofilament drawn at the exit of bushing Monofilament drawn from industrial strand Industrial strand constituted of a large number of filaments E-glass 3,400 2,0002,400 1,2001,550 R-glass 4,400 3,600 1,7002,000
TABLE 2.5. Fracture characteristics of industrial glass strands, deduced from the characteristics measured on unidirectional fibre composites. Characteristics Tensile fracture stress Ultimate strain (MPa) (%) Verre E 2,4002,600 3.4 Verre R 3,0003,600 4
obtain a simultaneous and uniform loading of all the filaments which constitute the strands. When strands are incorporated in composite materials, the fibrematrix bonding induced by the sizing process ensures a fairly homogeneous distribution of the loading on the strand filaments. The values of fracture characteristics, deduced from the fracture properties of unidirectional composites are reported in Table 2.5. These values are quite close to those measured on monofilaments obtained from industrial strands (Table 2.4), and usually these values may be considered as being representative of the fracture properties of glass fibres. It has to be noted that glass fibres keep their mechanical properties up to quite high temperatures, of the order of 200 C for E-glass and 250 C for R-glass. These fibres are thus well adapted to reinforcing matrices that have high thermal properties.
2.5.1.5 Industrial Glass Fibre Products

2.5.1.5.1. Basic Fibres It has been reported in Subsection 2.5.1.2 that the sized filaments are assembled together without twisting at the exit of the bushing equipment to form a strand. The glass strands are only an intermediate step, and can then be processed into different industrial products used in the manufacturing of composite materials. Strands can be characterised by: the designation of the type of glass used: E, R, S, etc., the type of strand: continuous (C) or discontinuous (D), the nominal diameter of filaments, the linear density of the thread in tex unit.
34
For example, the strand might be referred as EC 9 34: continuous strand of Eglass fibre, with filament diameter of 9 m and strand linear density of 34 tex. This designation conforms to the ISO 2078 standard. 2.5.1.5.2. Milled Threads Milled fibres are produced by milling continuous strands. Fibre lengths are of the order of a few tenths of a millimetre, with a length-to-diameter ratio of about 10 to 40. Milled fibres are generally used as fillers for resin systems used in casting, encapsulating and tooling applications. Fibre loading of up to 40 % weight can be obtained. 2.5.1.5.3. Chopped Strands Chopped strands are obtained by cutting continuous strands. They are usually supplied in lengths of 3, 4.5, 5, 13, 25 and 50 mm. The chopped strands are used in the manufacture of dough moulding compounds and as reinforcement of thermoplastic resins. 2.5.1.5.4. Rovings and Yarns Continuous rovings are obtained by assembling parallel strands, without twisting, and winding into a spool or onto a tube to form a cheese (or cake) (Figure2.7). The designation of the roving indicates either the linear density or the number of basic strands. For example, for roving EC 10 2,400 (global designation) or roving EC 10 4060 (complete designation), EC 10 40 denotes the basic strands, 60 gives the number of basic strands assembled without twisting, 2,400 indicates the global linear density in tex of the roving. The different types of rovings are produced to meet the needs of the different processes of composite moulding. Rovings differ by: the number of basic strands: 2, 8, 15, 30, 60 for example; the global linear density: 600, 1,200, 2,400, 4,800, 9,600 tex.
FIGURE 2.7. Different forms of roving supplies: (1) Spools, (2) Roving Cheeses or Cakes wound on tubes (Vetrotex documentation).
35
Applications of rovings are varied: chopping for fabrication of mats, for spray layup; weaving for making surfacing cloths or preforms; continuous impregnation, as filament winding, pultrusion; etc. To obtain rovings, the strands are assembled without twisting. The strands can also be twisted into yarns and next woven into weave fabrics. The twist may be of about 20 to 40 turns per metre and leads to a uniform cloth to be woven. 2.5.1.5.5. Curly Rovings Curly rovings (Figures 2.8 and 2.9) are made by assembling curly glass strands. The structure of curly rovings provides the ability to have some reinforcement in the direction transverse (or close to transverse) to the direction of strands. Curly rovings are used essentially in weaving heavy cloth. Such cloth allows an improvement in the behaviour of composite materials subjected to interlamina shear between composite layers. 2.5.1.5.6. Chopped Fibre Mat Chopped fibre mats are constituted of chopped strands or rovings randomly distributed in a plane. A typical construction is shown in Figures 2.10 et 2.11. Chopped fibre mats differ by the characteristics of the chopped strands or rovings (type of glass, fibre diameter, linear density, etc.) and by the characteristics of the binder used to hold the fibres together. Usually the binders are chemical ones, in the forms of liquids or powders. A liquid binder generally imparts greater drapability of the mat than a solid binder. High binder contents are used to prevent the fibres to flow with the resin during moulding processes. Mechanical bonding is also used with lightweight roving. The chopped strands or rovings are needled into the others. Such mats show a good drapability and fast wet out. The length of fibres is usually 50 mm. The weight per unit area is currently 300, 450 and 600 g/m2. However, for particular uses it is possible to obtain products with weights less than 300 g/m2 or greater than 600 g/m2, for example 900, 1,200 g/m2. 2.5.1.5.7. Continuous Fibre Mat Continuous fibre mats (Figures 2.12 et 2.13) are rather similar to chopped fibre mats. They have a better deformability and thus improved drapability in the composite processes. Usually, continuous fibre mats are made from strands having a low linear density of the order of 25 tex. 2.5.1.5.8. Woven Glass Fabrics (Figures 2.14 to 2.18) The different types of woven fabrics have been presented (Subsection 2.4.3) in the general case. Woven glass fabrics are supplied in the forms of rolls of two sizes: large width (Figures 2.14 to 2.16), close to a metre for strand cloths, or up to about 3 m in the case of roving or yarn cloths; small width in the form of ribbons (Figure 2.17).
36
FIGURE 2.8. Curly roving: Commercial forms (Documentation Vetrotex).
FIGURE 2.9. Curly roving: detailed view (Documentation Vetrotex).
37
FIGURE 2.10. Commercial supply of chopped fibre mat (Documentation Vetrotex).
FIGURE 2.11. Detailed view of chopped fibre mat (Documentation Vetrotex).
38
FIGURE 2.12. Commercial supply of continuous fibre mat (Documentation Vetrotex).
FIGURE 2.13. Detailed view of continuous fibre mat (Documentation Vetrotex).
39
FIGURE 2.14. Examples of three glass plain weaves, processes from rovings of different linear density: 2,400, 1,200 and 320 tex (Documentation Vetrotex).
FIGURE 2.15. Plain weave fabric, made with glass strands (Documentation Vetrotex).
FIGURE 2.16. Twill weave (or serge) (Documentation Vetrotex).
40
FIGURE 2.17. Unidirectional roving weave fabrics: (1) unidirectional cloth in the weft direction, (2) unidirectional cloths in the warp direction (Documentation Vetrotex).
FIGURE 2.18. Complex constituted of unidirectional roving cloth and mat, with chemical bonding (Documentation Vetrotex).
41
The weights of glass cloths are functions of the characteristics of threads and weaving parameters. They are currently in the ranges: from 50 to 500 g/m2 for strand cloths, from 150 to 1,000 g/m2 for roving cloths and yarns cloths.. Moreover, there exists complex arrangements (Figure 2.18) which combine mat with weave cloth. The principal application of these products is the reinforcement of structures of plane forms or low curvature. When such structures are subjected to bending loading, the unidirectional cloth is put in the tensile area and the mat in the compressive area, according to their respective properties suited to the two types of loading.
2.5.2 Carbon Fibres

2.5.2.1 General Features
Graphite has a structure where the carbon atoms are distributed in parallel crystallographic planes (Figure 2.19). Carbon atoms in the planes are arranged according to a hexagonal structure in such a way that a carbon atom of a layer is projected at the centre of a hexagon of neighbouring layers. The bonds between carbon atoms in neighbouring planes are weak, giving the graphite good thermal and electrical conduction properties. In contrast, the bonds between atoms in the same plane are strong, providing high mechanical properties in directions parallel to the crystallographic planes. Theoretical considerations on the bonds between carbon atoms predict a Youngs modulus of 1,200 GPa and a tensile fracture stress of 20,000 MPa. Furthermore, the low density (less than 2,000 kg/m3) leads
layer 3
layer 2 3.35
layer 1
1.42
FIGURE 2.19. Structure of graphite.
42
to remarkably high theoretical specific mechanical properties. These considerations explain the extensive developments of fabrication processes for elaborating carbon fibres with directions of the crystallographic planes which is as parallel as possible to the fibre axes. Industrial fibres do not reach the theoretical mechanical values because imperfections in the crystalline structure. However, mechanical characteristics in fibre direction remain high and can reach the order of 700 GPa for the Youngs modulus and 4,000 MPa for the tensile fracture stress. Mechanical properties in the transverse direction of fibres are lower.
2.5.2.2 Manufacturing Carbon Fibres

2.5.2.2.1. From Acrylic Fibres Carbon fibres are elaborated through thermal transformations of various organic precursor fibres. The initial developments used rayon as the precursor material and resulted in relatively low performance fibres. Actually, carbon fibres are prepared from polyacrylonitrile (PAN) precursors, using controlled oxidation and carbonization of the precursor fibres at high temperatures. These acrylic fibres are known in commercial names as: crylor, courtelle, dralon, orlon, etc. The mechanical properties of the final carbon fibres depend on the quality of the precursor. The elaboration principle is to submit the acrylic fibres to a thermal decomposition, without melting, and leading next to a carbonization process which keeps the initial structure of the precursor fibres. The actual processes used strands of acrylic fibres assembled without twisting (currently 500, 1,000, 6,000, 10,000, etc. filaments), and subject fibres to four successive treatments: oxidation, carbonization, graphitization and surface treatment (Figure 2.20). Oxidation. The object of the oxidation phase is to artificially suppress the melting point of acrylic fibres. This operation is carried out by heating the fibres to around 300 C in an oxygen atmosphere. This stage produces an oxidation that leads to
inert gas
GRAPHITIZATION HNO3 OXIDATION 300 C CARBONIZATION 2,600 C 2 possible routes OXIDATION HM or UHM fibres
HS fibres
oxygen acrylic fibres
inert gas
HNO3
FIGURE 2.20. Process of carbon fibre production from PAN filaments.
43
the formation of a cross-linked molecule structure in fibres:

CN CH CH2 2 PAN molecules CH2 CH CN CH CN CN CN CH2 CH CH CH CH CH CN CH CH2 + O2 CH CN CH CH O CN CH CH O
300 C
Carbonization. The second stage consists of progressively heating the crosslinked fibres from 300 C to about 1,100 C in a furnace with inert atmosphere. Water and cyanhydric acid are then eliminated, and only the carbon atoms are preserved in the chain:
CN CH CH O CH CH CN CH O CH CH CN CN CH CH O C CH C C C C C C C C + H20 + 4HCN
1,100 C
The fibres obtained after this stage have good mechanical properties and can be used after surface treatment (Figure 2.20). The fibres are then called HS carbon fibres (high-strength carbon fibres). Graphitization. The graphitization process is used for obtaining fibres which have a high Youngs modulus. This process consists of carrying out after carbonization a pyrolysis of the fibres in an inert atmosphere at temperatures up to 2,600 C. The graphitization stage causes a high degree of orientation of the hexagonal structure of carbon atoms along the axis of the fibres, which leads to an increase in the Youngs modulus. However, simultaneously with this orientation, defects are induced in the carbon structures, implying a decrease in the fibre strength. According to the level of graphitization, the high strength fibres are converted into HM (high modulus) fibres or UHM (ultra high modulus) fibres. Surface treatment. The last stage of fibre elaboration consists of a surface treatment by oxidation in an acid atmosphere (nitric or sulfuric). The object of this treatment is to increase the roughness of the filament surface and to improve bonding between the matrix resin and the fibres.
44
2.5.2.2.2. From Pitch Precursor Since the mid 1970s, processes of carbon fibre elaboration were developed from pitch as precursor. Pitch is a residue of oil refining. In these processes, the pitch precursor is first converted into a mesophase (intermediate between liquid and crystal phase) by heating the pitch to about 350 C to 450 C. Then, the mesophase is transformed into a filamentary form by spinning or extrusion, which improves the molecular orientation. As in the case of PAN process, the filaments obtained are oxidized and carbonized, and next pyrolized above 2,000 C to obtain high modulus fibres. Carbon fibres produced by this process have significant production advantages: a high weight efficiency: a fibre/precursor ratio of about 75 to 90 % (50 % for the PAN process, a faster speed of graphitization, a cheaper precursor material. The development of this technique would allow carbon fibres to reach the commercial markets, such as the automobile market, with a significant decrease in fibre cost price compared with fibres obtained from the PAN Precursor.
2.5.2.3 Mechanical Properties of Carbon Fibres

Carbon fibres have very good mechanical properties, associated with low density (usually less than 2,000 kg/m3). Table 2.6 compares the properties of carbon fibres with those of E-glass fibres. Furthermore, it must be noted that carbon fibres have an excellent thermal behaviour in a non-oxidising atmosphere. Their mechanical properties are kept up to around 1,500 C. This property has led to developing composite materials with carbon fibres and carbon matrix with a high temperature behaviour, used in rocket nozzles, brake blocks (trucks, Formula 1 cars, airplanes), oven elements, etc. These materials, covered with an antioxidant atmosphere, are also used for applications in oxidant atmosphere in the space area, such as thrust bearings or tiles.
2.5.2.4 Industrial Products

Like glass fibres, carbon fibres can be obtained in a number of commercial forms. Carbon fibres are manufactured as continuous filaments assembled together without twisting into tows (equivalent to rovings in glass terms) consisting of 500, 1,000, 3,000, 6,000, 10,000, etc., monofilaments, depending on manufacturer. These tows can then be used directly for fabrication processes as pultrusion or filament winding. They can also be used for the manufacture of woven carbon fabrics, mats, multidirectional fabrics, ribbons, preforms, etc. Carbon strands can be chopped to lengths of a few millimetres and incorporated in resin matrix.
45
TABLE 2.6. Properties of carbon fibres, compared with those of E-glass fibres. Properties Density (kg/m3) Diameter (m) Youngs modulus Ef (GPa) Specific modulus Ef / (MNm/kg) Tensile fracture stress fu (MPa) Specific fracture stress fu/ (kNm/kg) Cost price referred to E-glass fibres E-glass HS Carbon 1,750 57 230 130 3,000 4,000 1,710 2,290 1015 HM Carbon 1,810 57 400 210 2,800 1,550 3050 UHM Carbon 1,950 57 600 310 2,000 1,030 200400 HM Carbon (pitch) 2,000 12 280 140 2,000 2,400 1,000 1, 200 50100
2,600 1020 73 28 3,400 1,300 1*
1* 4 $/kg en 2006.
HM (pitch): fibres obtained from pitch.
2.5.3 Aramid Fibres with High Mechanical Properties

2.5.3.1 General Elements
Aramid fibres with high mechanical properties are usually known as Kevlar fibres, the trade name of fibres developed by Dupont de Nemours (USA) and introduced into the composite market in 1972. Other groups also produce aramid fibres, in particular the AKZO group (Germany and Holland) which produces aramid fibres under the trade name of Twaron fibres, and the Japanese group Teijin Ltd which supplies Technora fibres. Aramid fibres are aromatic polyamide fibres constituted of benzene kernels linked by CO and HN groups:
H N H N O C O C
Fibres are produced by extrusion and spinning processes. A solution of the polyamide polymer is held at low temperatures (between 50 C and 80 C and then extruded at a temperature of about 200 C. Aramid fibres are then drawn and conditioned to improve their mechanical properties.
46
2.5.3.2 Properties
Typical mechanical properties of aramid fibres are reported in Table 2.7 for different types of monofilaments. For multifilament strands, the mechanical characteristics are usually lower. The table shows a high specific tensile fracture stress, of the same order of height as the HR carbon fibres (Table 2.6), at a cost price 4 to 6 times less. Nevertheless, their use is limited by some weaknesses of composite materials with aramid fibres, as: low resistances to compression, bending, buckling; sensitivity to interlaminar shear. These weaknesses are usually associated to a poor interfacial bonding between fibres and matrix. Different fibre treatments have been developed for improving the interfacial bonding and the delamination resistance of the aramid fibre composites. However, the poor interlaminar bond between fibres and matrix leads to high impact resistance and damage tolerance of composites. This results from the absorption of energy by widespread delamination and splitting of fibres.
2.5.3.3 Industrial Uses

Aramid fibres are available in the form of strands, roving and woven fabrics. One of their industrial applications was to replace steel armatures in tires, belts and pipes. Due to their lightness and good resistance to shock, impact and damage tolerance, aramid fibres are used in sport applications (skis, tennis rackets, etc.) and for ballistic applications (armour plating, armoured jackets, helmets, gloves, etc.). Another application of the aramid fibres is the manufacture of high strength ropes and cables. Hybrid composites of carbon fibres (or glass fibres) and aramid fibres are used to improve the impact resistance of carbon (or glass) fibre composite materials.
TABLE 2.7. Mechanical properties of aramid filaments. Characteristics Density (kg/m3) Diameter (m) Youngs modulus Ef (GPa) Specific modulus Ef / (MNm/kg) Tensile fracture stress fu (MPa) Specific fracture stress fu / (kNm/kg) Ultimate elongation (%) Kevlar 29 1,440 12 60 42 3,000 2,080 3.6 Kevlar 49 1,450 12 120 83 3,000 2,070 1.9 Kevlar 149 1,470 12 160 110 2,400 1,630 1.5 Twaron Technora
1,440 12 60 42 2,600 1,800 3
1,390 12 90 65 2,800 2,010 4
47
2.5.4 Ceramic Fibres

2.5.4.1 General Features
Different fibres constituted of refractory or ceramic materials (carbides, borides, nitrides, etc.) can be produced by chemical deposition of vapour phase onto a thread support. In practice, fibres obtained by this process, which are the object of production, are: boron (B) fibres, boron carbide (B4C) fibres, silicon carbide (SiC) fibres, boron-silicon carbide (called BorSiC) fibres. In fact, the fibres are large diameter threads (diameter of the order of 100 m), made of a core (the support thread) of tungsten or carbon of a diameter of around 10 to 20 m, covered with: a layer of boron about 40 m thick (boron fibres), a layer of boron about 40 m thick and a layer of boron carbide 4 m thick (B-B4C fibres), a layer of silicon carbide (SiC fibres), a layer of boron and a layer of silicon carbide (BorSiC fibres).
2.5.4.2 Mechanical Properties and Applications

The mechanical properties of the different ceramic fibres are quite close, see Table 2.8 for examples. These characteristics are kept at temperatures up to 500 to 1,000 C. Due to their high cost, the use of these fibres is limited. Actually, SiC and BorSiC fibres are used essentially with metallic (aluminium, for example) or ceramic matrices, leading to very expensive composite materials used for blades in compressors, turbines, etc.
TABLE 2.8. Mechanical properties of fibres of boron, boron-boron carbide and silicon carbide. Characteristics Density Diameter Youngs modulus Specific modulus Tensile fracture stress Specific fracture stress Boron 2,600 100150 430 165 3,800 1,460 Boron + B4C 2,600 100150 430 165 4,000 1,540 SiC 3,000 100150 410 140 3,00 1,300
(kg/m3)
(m) Ef (GPa) Ef / (MNm/kg)
fu (MPa) fu / (kNm/kg)
48
Boron fibres and boron-silicon carbide fibres are available in the form of: continuous strands made of parallel filaments, peimpregnated fibrous yarns or coated filaments in spools for filament windding and pultrusion, woven fabrics and preforms. These fibres associated to epoxide matrices, with which adhesion is very good, or with other thermoplastic or themosetting matrices, are used in the sport and leisure domains.
2.5.4.3 Other Types of Ceramic Fibres

Other types of ceramic fibres, as silicon carbide (SiC) or silicon carbidetitanium (SiCTi) fibres, are obtained by the precursor process similar to the process used for elaborating carbon fibres from PAN precursor. In contrast to fibres considered in the previous subsection, the precursor process leads to ceramic fibres with small diameters. Typical mechanical properties of these fibres are reported in Table 2.9. Also, these fibres have a good stability in their properties up to temperatures of 1,200 to 1,600 C. Some fibres with an alumina base are also developed, as alumina (Al2O3) fibres, aluminosilicate (Al2O3-SiO2) fibres and borosilicoaluminate (Al2O3-SiO2B2O3) fibres. These fibres are obtained by a sol-gel route process, and the mechanical characteristics are reported in Table 2.10.
2.5.5 Thermostable Synthetic Fibres

Thermostable synthetic fibres are organic fibres obtained by synthesis processes. These fibres keep their mechanical properties at high temperatures. Associated with thermostable resins, they allow materials to be obtained with mechanical properties which are maintained with temperature. The mechanical properties of these fibres are, however, much lower than the usual fibres. They are used in electrical and thermal insulators and for thermal protection: noses and leading edge structures of shuttles and missiles, space reentry cones, etc.
2.5.6 Other Fibres

Various other fibres are used for particular applications. These fibres generally have low modulus and strength, except the metal fibres. Usually, these fibres are used for obtaining: low cost products, thermal or electrical insulation products, products with good thermal or electrical conductivity, products with high acoustic absorption, etc.
49
TABLE 2.9. Mechanical properties of fibres of SiC, SiNC and SiCTi. Characteristics Density Youngs modulus Specific modulus Tensile fracture stress Specific fracture stress SiC (kg/m3) (GPa) (MNm/kg) (MPa) 2,550 180200 7080 3,000 1,200 SiNC 2,350 170 70 2,400 1,000 SiCTi 2,400 200 85 3,000 1,250
Ef E f /
fu
fu / (kNm/kg)
TABLEAU 2.10. Mechanical properties of ceramic fibres with an alumina base. Alumina Al2O3 (kg/m3) (GPa) 3,400 3,950 300390 90100 1,500 2,000 440500 Aluminosilicate Al2O3, SiO2 3,100 3,200 190250 6080 2,100 2,200 685 Borosilicoaluminate Al2O3, SiO2, B2O3 2,700 3,100 150200 5565 1,700 1,800 580630
Characteristics
Density Youngs modulus Specific modulus Tensile fracture stress Specific fracture stress
Ef
Ef / (MNm/kg)
fu
(MPa)
fu / (kNm/kg)
Among these fibres are: 1. Fibres of vegetable origin, as sisal, jute, or flax, used as woven cloths, in yarn form for filament winding or in dough moulding compounds. 2. Synthetic fibres, as polyester (tergal, dacron, etc.) fibres, polyamide fibres, polyethylene fibres, polypropylene fibres, etc. 3. Metal fibres, as steel fibres, copper fibres, aluminium fibres, used with metal matrices for thermal, electrical or thermomechanical applications.
CHAPTER 3
Architecture of Composite Materials
3.1 INTRODUCTION
Usually, the elaboration of structure of composite materials is carried out by moulding processes in which the structures are moulded by successive layers constituted of matrix and reinforcement. This general technique, usually called lamination, leads to forming laminated materials or laminates. Furthermore, laminates can be associated with light materials to obtain sandwich materials. Moulding processes are considered in References 6 and 7. Only, the architecture of composite materials deduced from the considerations of the moulding processes will be developed hereafter in this chapter. Architecture of composite materials constitutes an introduction to the study of the mechanical behaviour of composite materials which will be developed next along this book. So, the various moulding processes show the general feature of composite structures to be designed in the form of plates and shells by lamination of successsive layers. This design justifies the importance that will, hereafter, be placed in the analysis of composite materials in the form of plates of one single layer or of several distinct layers. Next, shells can be modelled as a set of plates, and their study derived from the analysis of plates.
3.2 LAMINATES 3.2.1 Introduction

Laminates are constituted of successive layers (Figure 3.1) (sometimes called plies) of reinforcements (strands, rovings, mats, cloths, etc.) impregnated with resins. The layers differ by the constituents, the layer orientations, etc. In this section, we consider general elements on the various types of laminates.
3.2 Laminates
51
layers
laminate
FIGURE 3.1. Laminate constitution.
3.2.2 Laminates Constituted of Unidirectional layers

3.2.2.1 Laminate designation
Laminates with unidirectional fibres constitute a basic laminate from which, in theory, the analysis of every other type of laminate can be deduced. These unidirectional laminates are constituted of parallel strands or unidirectional cloth embedded in a matrix. Several unidirectional layers can then be stacked (Figure 3.2) in a specified sequence of different fibre orientations to obtain a laminate that will fit the mechanical properties required.
laminate: 45 0 45 90 90 30 designation: x 30 90 90 45 0 45 z
[30 / 902 / 45 / 0 / 45]
FIGURE 3.2. Stacking sequence in a laminate.
52
Chapitre 3 Architecture of Composite Materials
The identification of a laminate can be obtained by using the following orientation code: 1. Each layer is identified by the orientation in degrees between the fibre direction and the x reference axis (Figure 3.2). 2. Successive layers are separated by /, if the layers orientations are different. 3. Successive layers with the same orientation are indicated by using a numerical index. 4. The layers are successively designed going from one face to the other. Brackets indicate the beginning and the end of the code. The designation depends on the axis system chosen. An example is reported in Figure 3.2.
3.2.2.2 Positive and Negative Angles

When layers are oriented at angles equal in magnitude but with opposite signs, the + or sign is used. The designation for positive or negative angles depends on the axis system chosen: the signs are reverse when the z direction is reversed (Figure 2.3). The laminate designation is [30/90/45/0/45] for the upwards orienttation of z axis, and [30/90/45/0/45] for the downwards orientation. The mechanical behaviour of laminate is different for the two stacking sequences.
30
90
45
45
y x 30 90 45 0 45 y x z
FIGURE 3.3. Sign notation for the stacking sequence of a laminate.
3.2 Laminates
53
3.2.2.3 Examples
Some examples of laminates and designations are given hereafter. Laminate 0 30 30 45 45 Designation
[ 45 / 30 / 0 ]
30 60 60 0 45
[45 / 0 / 602 / 30]
0 45 45 45 45
[452 / 452 / 0]
0 45 45 45 45 45 45
[ 45 / 0] or [45 / 452 / 452 / 45 / 0 ]
3.2.2.4 Symmetric Laminates

A laminate is symmetric if its midplane is the plane of symmetry. In this case, only half of the laminate is necessary for the laminate designation.
54
If a symmetric laminate has an even number of layers, its designation starts from a face and finishes at the midplane. An index S indicates that the laminate is symmetric. For example: 90 45 45 0 0 45 45 90 If the laminate has an odd number of layers, the designation is similar to the preceeding one, but the central layer is overlined. For example: 90 45 45
[90 / 452 / 0]S
[90 / 452 / 0 ]S
0 45 45 90
3.2.2.5 Laminates with Sequences

Repeating sequences can be indicated by a subscript showing the number of times that a sequence is successively repeated. For example: 0 45 90 0 45 90 90 45 0 90 45 0
[(0 / 45 / 90)2]S or [0 / 45 / 90]2S
3.2 Laminates
55
0 45 90 0 45 90 60 30 60 30 60 30
[(30 / 60)3 (90 / 45 / 0)2]
The particular stacking sequence [0/90]n is referred to as cross-ply laminate.
3.2.2.6 Hybrid Laminates

Hybrid laminates are constituted of successive layers made of fibres of different natures: glass, fibres, carbon fibres, Kevlar fibres, etc. It will then be necessary to indicate the nature of fibres in the designation of hybrid laminates. For example: 0 45 45 90 90 45 45 0 G C C C C C C G
[0G / 45C / 90C]S
0 0 45 90 45 0 0
K K G G G C C
[02C / (45 / 90 )SG / 02K]
where G stands for glass, C for carbon and K for Kevlar.
56
3.2.3 Laminate Structure

In the general case, the reinforcement in each layer will be of various kinds: strands, rovings, mats, cloths, glass fibres, carbon fibres, Kevlar fibres, etc. Each layer must then be designated by the nature of fibres (glass, carbon, Kevlar, etc.) and the types of reinforcement (strand, mat, cloth, etc.), with the indication of the fibres proportion in the warp and weft directions. The choice of the nature and the stacking sequence of layers will depend on the use of the composite material, adapting the material to the loading conditions under consideration. For example: Unidirectional layers have good mechanical properties in the direction of fibres. Mats have low resistance to tension and have a good resistance in compression. A cross-ply laminate is sensitive to interlaminar delamination. A lamination with at least three fibre directions (0, 90 and 45) are necessary if quasi-isotropy is required in the plane of the laminate. Etc. Also, an important feature lies in the fact that a symmetric lamination usually avoids any warping of the laminate after demoulding. For example, for a beam or a plate submitted to bending loading, unidirectional layers can be put in the tension zone and mat layer in the compression zone (Figure 3.4).
3.2.4 Hybrid Laminates

Hybrid laminates allow to obtain the optimum properties by using the different properties of fibres of glass, carbon or Kevlar. Among the different hybrid laminates, it is distinguished: the interply hybrid laminates, constituted of layers each with different nature of fibres, the intraply hybrids, made by a sequence of layers, each layer being constituted with reinforcements of different natures. Also, particular layers can be interleaved between layers, as viscoelastic layers, for example, to increase the damping properties of the laminates [3].
mat
unidirectional
FIGURE 3.4. Beam bending.
3.3 Sandwich Composites
57
3.3 SANDWICH COMPOSITES

The principle of sandwich construction consists in coating to a core on both sides two sheets, called skins. The core is made of a light material or structure which must have good properties when submitted to transverse compression. The skins must have good properties in tension. The objective of sandwich concept is to obtain a material combining lightness and high flexural stiffness. Usually, the choice of sandwich constituents is made with the initial objective of having a minimal weight, next taking into account the conditions of use (thermal behaviour, corrosion resistance, cost, etc.). The materials most frequently used are: for solid cores (Figure 3.5) : the balsa or cellular wood; various cellular foams; resins filled with hollow glass microspheres, called syntactic foams; etc. ; for hollow cores (Figure 3.6), essentially honeycombs and profiles of light materials: light metal alloys; kraft paper (coated or no with resin); polyamide paper, type Nomex paper; etc.; Mixed cores can be used.
laminates
foam core
balsa core
grain of balsa
laminate
FIGURE 3.5. Sandwich materials with solid cores.
58
honeycomb core
corrugated core
FIGURE 3.6. Sandwich materials with hollow cores.
The sandwich skins are most frequently constituted of laminates with glass fibres, carbon fibres or Kevlar fibres. Light alloy sheets are also used. In order that sandwich constructions are efficiency, it is necessary to have a good bonding between the core and the skins so that the mechanical loading can be transmitted between core and skins. The bonding is obtained by using resin systems for skins compatible with the core materials or by interleaving particular interface layer between core and skins.
3.4 OTHER COMPOSITE MATERIALS

The other composite materials can be schematically classified into reinforced plastics and volume composites. The analysis of the mechanical behaviour developed in the present book will consider only laminate and sandwich materials.
3.4.1 Reinforced Plastics

These materials are constituted of plastic matrices in which reinforcements are incorporated in the form of: short fibres, solid or hollow microspheres, powders, as metal powder, graphite powder, etc.
3.5 Consequences on the Mechanical Behaviour Analysis of Composite Materials
59
These reinforcements usually increase the elasticity modulus by a factor of about two to five. The mechanical behaviour of these materials can be homogenized and the mechanical behaviour analysis is then reduced to that of a usual isotropic material.
3.4.2 Volume Composites

Volume composites were introduced for some particular needs of aeronautics. They are made from volume weavings (Subsection 2.4.4.2). These materials allow to achieve high mechanical properties with a behaviour that is practically isotropic in all the directions. However, the elaboration of these materials is very expensive.
3.5 CONSEQUENCES ON THE MECHANICAL BEHAVIOUR ANALYSIS OF COMPOSITE MATERIALS

The study of the manufacture of composite materials has shown the importance of laminates and sandwich materials. The architecture of these materials now indicates the guiding features of the analysis of their mechanical behaviour. This analysis will be developed according two general steps: 1. The study of the mechanichal behaviour of each layer as function of the layer constituents: matrix and fibres. This study is quite often called as microanalysis of the composite materials. We prefer the analysis at the constituent level. This analysis will be developed in Part III. 2. The study of the global behaviour of laminate or sandwich constituted of several layers. This global behaviour is generally called the macroscopic behaviour of composite. We will refer to as laminate behaviour or sandwich behaviour. This study will be considered in Part IV. When these two studies have been implemented, the mechanical behaviour of structures constituted of composite materials can next be analysing by adapting the usual tools of structure analysis to the study of composite structures. The analysis of simple structures (beams and plates) can usually be achieved by analytical methods (Part V). The study of complex composite structures requires the use of finite element method (Chapter 26 of Part V). The steps of the analysis of mechanical behaviour of composite material structures are shown schematically in Figure 3.7.
60
laminate laminate behaviour layer analysis at constituent level
F3
F1
F2
STRUCTURE displacement conditions imposed on the boundary
F1 F2 F3 loads imposed
For describing the mechanical behaviour of composite structure it is necessary to know the stress field and displacement field at every point of the structure
Study Process Analysis at Constituent Level of a Layer Laminate or Sandwich Behaviour Analysis of the Composite Structure
Analysis at Constituent Level Analysis of the elastic properties and fracture behaviour of layer as function of constituents
Laminate or Sandwich Behaviour Study of the elastic properties and fracture behaviour of a laminate or sandwich material as function of layers
FIGURE 3.7. Analysis of the mechanical behaviour of a composite material structure.
Part II
Fundamental Elements on the Mechanical Behaviour of Materials and Structures
To develop the concepts of the mechanical analysis of composite materials and structures, it is necessary to be familiar with the basic theory of the mechanics of deformable structures. A synthesis of the basic elements is presented in this part, considering the classical concepts: stresses (Chapter 5), strains (Chapter 6), the elastic behaviour of materials (Chapter 7) and the fundamental elements of the mechanical analysis of deformable structures (Chapter 8).
CHAPTER 4
Mathematical Basics
The strain state and stress state at a point of a deformable solid submitted to a mechanical loading are characterised by 3 3 square matrices which depend on the orientation of reference system. In this chapter, basic properties on matrices and transformations are recalled.
4.1 TRANSFORMATION OF BASIS VECTORS 4.1.1 General Expression

The orientation of a geometric space (S) with respect to a reference system (R) (Figure 4.1) is characterised by the basis vectors ( e ) = ( e1, e2 , e3 ) of a system of axes (1, 2, 3) linked to the space (S), ( e ) being a basis of the vector space 3 . Any transformation of this basis system is characterised by a transformation ( ' ) ' ' ' matrix A = aij , allowing to express the new basis vectors e = ( e1, e2 , e3 ) as
functions of the previous ones as: ' = a11e1 + a12e2 + a13 e3 , e1 ' = a21e1 + a22e2 + a23e3 , e2 ' = a31e1 + a32e2 + a33e3. e3 This relation can be expressed in the matrix form:
(4.1)
' e e 1 1 ' e2 = A e2 . ' e e2 3
(4.2)
where A is known as the transformation or rotation matrix. The inverted relation
64
Chapter 4 Mathematical Basics
3
e3
2
e2 e1
(S)
1 (R)
FIGURE 4.1. Orientation of a geometric space (S).
is written, by introducing the inverse matrix A1 of A, as:
e 1 e2 = A 1 e2
' e 1 ' e 2 . ' e 3
(4.3)
In the case of orthonormal bases, the transformation matrix is symmetric and unitary: its determinant is unity and its inverse is identical to the transposed matrix. The inverse relation is thus simplified as: ' e e 1 1 ' e2 = A t e2 , (4.4) ' e e2 3 where the matrix At is the transpose matrix of A.
4.1.2 Case of Rotation around an Axis

In the case of a rotation around the direction e3 (Figure 4.2), the relations ' ' ' between basis vectors ( e1 , e2 , e3 ) and ( e1, e2 , e3 ) are written as:
e1' = e1 cos + e2 sin , ' e2 = e1 sin + e2 cos , ' e3 = e3 .
(4.5)
Whence, the transformation matrix is:
4.2 Second-Order Tensors
65
2
e 2 e2 e1
e3 e1
FIGURE 4.2. Rotation around the 3-direction.
cos A= sin 0
sin cos 0 sin cos 0
0 . 0 1 0 0 . 1
(4.6)
The inverse transformation matrix is: cos t A = sin 0
(4.7)
Whence, the inverse relation is: ' e1 = e1' cos e2 sin , ' cos , e2 = e1' sin + e2 ' . e3 = e3
(4.8)
4.2 SECOND-ORDER TENSORS 4.2.1 Introduction

The notion of tensor is necessary to establish the relation between physical effects and causes in anisotropic media. In such media, a cause applied in one direction in general produces an effect oriented in another direction. Physical phenomena are thus described by introducing tensors. Physical quantities which do not depend on the direction of measurement and that are characterised by a single number are described by scalars which are
66
tensors of zero-order. Vectors are first-order tensors. They describe physical quantities characterised by one number and one direction. More generally, complex physical quantities are described by tensors of order greater that one. In a general way, a tensor can be defined as establishing a linear relation between two tensors of lower order. In particular, a second-order tensor, defined over the vector space 3 , can be considered as a linear operator which establishes a correspondence between every vector X of the vector space 3 and a vector Y of 3 . This tensor is represented in the basis ( e ) by a table of elements Tkl called the components of the tensor in the basis ( e ) . This table consists of nine numbers (k, l = 1, 2, 3). In basis ( e' ) , this same tensor is represented by a table of elements related to the preceding ones by the expression: Tij
= aik a jlTkl , Tij
i, j, k , l = 1, 2, 3,
(4.9)
where the terms aik and ajl are the elements of the transformation matrix. The relation uses the summation convention under which a repeated subscript implies a summation. So, Relation (4.9) is equivalent to: = ai1a j1T11 + ai1a j 2T12 + + ai 3 a j 3T33 , Tij
i, j = 1, 2, 3.
(4.10)
Expression (4.9) can be considered as the defining relation of a second-order tensor. In fact, there is an identity between the components of second-order tensors and the elements of square matrices. In the case where the vector space reference is the space 3 , the components of second-order tensors can therefore be written in the following matrix form: T11 T12 T13 T = T21 T22 T23 . (4.11) T T T 31 32 33
4.2.2 Change of Reference System

In the case of direct orthonormal bases, Relation (4.9) of transformation can be expressed as: = aik Tkl alk . Tij (4.12) Taking account of the elements previously considered, this relation can be expressed in a matrix form as:
T = A T A t ,
(4.13)
. The inverted relation is: where T is the matrix Tij

T = A t T A .
(4.14)
4.2 Second-Order Tensors
67
4.2.3 Matrix Diagonalization: Eigenvectors and Eigenvalues

A nonzero vector u is an eigenvector of a tensor T if the vector Tu is collinear with u . Every vector collinear with an eigenvector is also an eigenvector. The direction characterised by an eigenvector is said a principal direction of the tensor. The eigenvector u must therefore satisfy the relation: (4.15) (T I) u = 0 ,
where I is the identity tensor of second order, that is the identity 3 3 matrix. This relation is written in an orthonormal basis ( e ) as:
(Tij ij ) u j = 0 ,
introducing the Kronecker symbol defined by:
(4.16)
ij =
As the vector u is different from the null vector, this system has solutions if:
1 if i = j , 0 if i j.
(4.17)
det ( T I ) = 0 .
(4.18)
This equation generally has three distinct roots: (1) , (2) , (3) , called the eigen-
values or principal values. To each eigenvalue ( k ) there corresponds an eigen vector u ( k ) the components of which satisfy the three non-independent equations:
Tij u (jk ) = ( k )ui( k ) .
(4.19)
In the case of symmetric tensors, we have the two fundamental properties: (1) the eigenvalues are real; (2) the principal directions are orthogonal. In the basis of eigenvectors, further assumed to be normed ui( k ) = ik , Relation (4.19) is then written: jk = ( k ) ik , Tij (4.20) Thus: = ( j ) ij , Tij (4.21)
are the new components in the eigenvector basis. The table Tij is thus where Tij diagonal: (1) 0 0 Tij = 0 (2) (4.22) 0 . 0 0 (3)
In the principal basis, the elements of a second order tensor are written as a diagonal matrix.
68
4.2.4 Inversion of symmetric Three-Order Matrix

The inversion relations of a symmetric 3 3 matrix are quite simple to use. They are written in the form:
Aij , A=
2 A22 A33 A23 ( ), 1 2 = ( A11 A22 A12 A33 ),
= A11
A = A 1 = Aij 1 2 = A22 A11 A33 A13 ,
= A12 = A23
( A13 A23 A12 A33 ) , ( A12 A13 A23 A11 ) ,
(4.23)
= A13
( A12 A23 A13 A22 ) ,
2 2 2 = det A = A11 A22 A33 + 2 A12 A13 A23 A11 A23 A33 A12 . A22 A13
CHAPTER 5
Stresses
5.1 STRESS STATE IN A SOLID

The stress state at a point M of a solid subjected to a mechanical loading (Figure 5.1) allows us to characterise the mechanical actions applied to the material at this point.
5.1.1 Stress Tensor

At every point M of a continuous medium, the stress state is completely determined by knowing the stress tensor, denoted (M). This tensor is a symmetric second-order tensor, represented by the matrix:
11 12 13 (M ) = 12 22 23 , 13 23 33
(5.1)
with
21 = 12 ,
31 = 13 ,
32 = 23 .
Solid
FIGURE 5.1. Solid subjected to a mechanical loading.
70
Chapter 5 Stresses
5.1.2 Force Acting at a Point of a Surface Element

The mechanical action, exerted on a unit surface (considered to be infinitely small) with orientation n (Figure 5.2) by the material situated on the side of positive orientation defined by the vector n (unit vector), is a force the resultant t ( M , n ) of which, called stress vector, is given by the relation : t ( M , n ) = t = ( M )n . (5.2) The force dF acting on the surface element dS of the same orientation is therefore expressed as: dF = t d S . (5.3) Expression (5.2) can be written in matrix form by introducing: the matrix of the components of the stress vector t in the basis (e1 , e2 , e3 ) :
t1 [ t ] = t2 , t3
(5.4)
the matrix of the components of the normal vector n in the basis (e1 , e2 , e3 ) :
n1 [ n ] = n2 . n3
(5.5)
We obtain:
t1 11 12 13 n1 t = 2 12 22 23 n2 . t3 13 23 33 n3 Whence, the components of the vector t in the basis (e1 , e2 , e3 ) are:
(5.6)
t1 = 11n1 + 12 n2 + 13n3 , t2 = 12 n1 + 22 n2 + 23n3 , t3 = 13n1 + 23n2 + 33n3 ,
(5.7)
and the stress vector is:
t = t1e1 + t2 e2 + t3e3 .
n
(5.8)
M dS
FIGURE 5.2. Surface element at a point M.
5.2 Properties of the Stress Tensor
71
5.2 PROPERTIES OF THE STRESS TENSOR 5.2.1 Physical Interpretation of the Stress Components
Let us consider a surface element of which the direction of the normal is the same as the direction of one of the reference axes (Figure 5.3). Thus: n = e j with j = 1, 2, 3.
From Relation (5.7), the components of the stress tensor are: t1 = 1 j , t2 = 2 j , t3 = 3 j , with ji = ij . The quantities 1 j , 2 j , 3 j are thus respectively the components in the direc tions (e1, e2 , e3 ) of the stress vector on a surface with normal direction n = e j ( j = 1, 2 or 3) . For example, if n = e1 , the components are respectively (5.9)
11, 21 = 12 , 31 = 31 , and the stress vector is:
t ( M , e1 ) = 11e1 + 12 e2 + 13e3 .
(5.10)
Hence the diagram of Figure 5.4. It is usual to represent the components of the stress vectors acting on the faces of a cube with faces parallel to the reference axes. Two notations are used and are reported in Figure 5.5, according to the axis reference: (1, 2, 3) or (x, y, z).
n = ej M
FIGURE 5.3. Surface element normal to the direction e j of one of the reference directions.
31e3 = 13e3 21e2 = 12e2 11e1

FIGURE 5.4. Components of the stress vector on a surface with normal direction e1 .
72
Chapter 5 Stresses
33 13 13 11
1
zz 23 23 22
2
x
xz
yz
yz yy
y
xy
xx
12 12
xy xy
FIGURE 5.5. Components of stress vectors: (a) in the set of directions (1,2,3); (b) in the set of directions (x, y, z).
5.2.2 Normal and Tangential Components of the Stress Vector

The stress vector t is usually decomposed (Figure 5.6) into: a normal stress tn in the direction n , a tangential stress tt with its direction contained in the vector plane of the surface element. This tangential stress is usually called the shear stress. The stress vector is then written as: t = tn + tt . (5.11) The normal stress is expressed as: tn = tn n , with tn = n t , (5.12)
where tn is the normal component of the stress vector. Introducing Relation (5.2), this component is given by: tn = n [ ( M ) n ] . (5.13)
t
tt tn
n
M
FIGURE 5.6. Normal and tangential components of the stress vector.
5.2 Properties of the Stress Tensor
73
This relation can be rewritten, using matrix notation, in the form:

tn = n t ( M ) n ,
(5.14)
on introducing the row matrix n t = [ n1 n2 We obtain:
n3 ] , the transpose of the matrix n.
2 2 2 tn = 11n1 + 22 n2 + 33n3 + 2 12 n1n2 + 2 13n1n3 + 2 23n2 n3 .
(5.15)
If tn is positive, it is said that the material is subjected to a tension (or a traction) at point M in the direction n . If tn is negative, the material is subjected to a compression. The shear stress can be determined by the expression: (5.16) tt = n {[ ( M )n ] n} .
5.2.3 Principal Directions and Principal Stresses

Because the stress tensor is a second-order tensor, there exists at least one basis (Chapter 4), called the principal basis of the stress tensor, in which the matrix representing the stress tensor is diagonal. In this principal basis, the stress matrix is written as:
(1) 0 0 0 . ( M ) = 0 (2) 0 0 (3)
(5.17)
The three components (1) , (2) , (3) are the principal stresses. If the stress tensor is expressed in a non-principal basis, the principal stresses are the roots of the equation: ij ij = 0, det where ij = 1 if i = j, and ij = 0 if i j, (i, j = 1, 2, 3). (5.18)
In the case where the surface at the point M is normal to a principal direction (i ) e , we find that:
tni = (i ) , tti = 0, i = 1, 2, 3.
(5.19)
For a principal direction, the shear stress is zero. The principal directions are thus such that ( M )n is collinear with n .
74
Chapter 5 Stresses

In the case of a rotation around the direction e3 (Figure 4.2), the relations between the stresses ij expressed in the basis (e1 , e2 , e3 ) and the stresses ij ' expressed in the basis ( e1' , e2 , e3 ) are obtained by combining Relations (4.6), (4.7) and (4.13). Thus:
= 11 cos 2 + 2 12 sin cos + 22 sin 2 , 11 = ( 22 11 ) sin cos + 12 (cos 2 sin 2 ), 12 = 13 cos + 23 sin , 13 = 12 , 21 = 11 sin 2 2 12 sin cos + 22 cos 2 , 22 = 13 sin + 23 cos , 23
In the case where the initial basis (e1 , e2 , e3 ) is the principal basis (1) (2) (3) (e , e , e ) of the stress tensor, the preceding relations reduce to:
(5.20)
= 13 , 31
= 23 , 32
= 33 . 33
= (1) cos 2 + (2) sin 2 , 11 = ( (2) (1) ) sin cos , 12 = 0, 13 = 0, 23 = 12 , 21 = 0, 31 = 0, 32 = (1) sin 2 + (2) cos 2 , 22 = (3) . 33
(5.21)
5.3 PARTICULAR STRESS STATES 5.3.1 Spherical Tensor and Stress Deviator
It is always possible to decompose the stress tensor ( M ) at a point M into the sum of two tensors: one, S ( M ), called spherical tensor; other, D ( M ) , called the stress deviator . Thus, the stress tensor is expressed as:
( M ) = S ( M ) + D ( M ) .
(5.22)
These tensors are defined in the following way. First, the spherical tensor is expressed as: s (5.23) S ( M ) = I , 3
5.3 Particular Stress States
75
where I is the unit tensor and s the trace of ( M ) : s = tr ( M ) . Next, the tensor D ( M ) is a tensor with zero trace that has the same principal directions as ( M ) and for principal stresses: s s s (1) , (2) , (3) . (5.24) 3 3 3
5.3.2 Spherical Compression or Tension

A stress state at a point is one of spherical compression or tension, when the deviator is the zero tensor. So, in this case we have: s ( M ) = S ( M ) = I . 3 This stress state is therefore characterised by the following properties: the stress vector t ( M , n ) is collinear with n ; the principal stresses are equal: (5.25)
(1) = (2) = (3) =

for arbitrary n , we have:
s ; 3
(5.26)
s tn = , 3
tt = 0.
(5.27)
5.3.3 Uniaxial Tension or Compression

A stress state is a uniaxial tension or compression state, if and only if two of the principal stresses are zero. For example: (2) = (3) = 0 . We then have a uniaxial stress state in the direction e (1) . If (1) > 0 , the stress state is a uniaxial tension. If (1) < 0 , the stress state is a uniaxial compression. It is easily found that the stress matrix is not modified by rotation around the direction e (1) . Like other properties, we have: the stress vector t ( M , n ) is always collinear with e (1) ; if n is orthogonal to e (1) , then: t ( M , n ) = 0 .
5.3.4 Shear State

At a point M, a stress state is a simple (or pure) shear state, when one of the principal stresses is zero and the other two are opposite. Thus, the determinant
76
Chapter 5 Stresses
e (3)
dS M
u
e (2)
e (1)
n
FIGURE 5.7. Surface element orthogonal to the principal plane e (1) , e (2) .
and the trace of the stress matrix are zero. For example:
(1) = , (2) = , (3) = 0.

The stress matrix is then written as:
0 (M ) = 0 0 0 0 0 . 0
(5.28)
(5.29)
Let us look for the stresses in the principal plane, that is the stresses acting on a surface element with the normal direction contained in the principal plane (1) (2) (e , e ) (Figure 5.7). Let be the angle that the normal n makes with the direction e (1) : n = (cos , sin , 0) . The stress vector at point M in the direction n is given by the expression:
0 t ( M , n ) = 0 0 0 0 cos 0 sin . 0 0
Thus:
t ( M , n ) = ( cos , sin , 0) ,
(5.30)
where the components are expressed in the principal basis ( e (1) , e (2) , e (3) ) .
5.3 Particular Stress States
77
Let us now look for the components of the stress vector in the basis (3) (3) ( n, u , e ) (Figure 5.7) obtained by a rotation around the direction e . They are determined by the expression:
cos t ( M , n ) = sin 0 sin cos 0 0 cos (cos 2 sin 2 ) . (5.31) 0 sin = 2 sin cos 1 0 0
The components in the basis ( n , u , e (3) ) are thus: t ( M , n ) = ( cos 2 , sin 2 , 0) .

The normal components is:
tn = cos 2 ,
(5.32)
(5.33)
and the tangential component is:

tt = sin 2 .
(5.34)
These expressions show the following properties: 1. For:
and
tn = 0, tt = , tn = 0, tt = .
(5.35)
=3
and
(5.36)
These results are illustrated in Figure 5.8a. 2. For:
=0 =
2
and and

2 +
tn = , tt = 0, tn = , tt = 0,
(5.37) (5.38)
The results are illustrated in Figure 5.8b.
Lastly, it is found that the stress matrix, expressed in a basis ( n , u , e (3) ) obtained from the principal basis by a rotation around the direction e (3) , is:
cos 2 sin 2 0
sin 2 cos 2 0
0 0 . 0
(5.39)
78
Chapter 5 Stresses
e (2)
e (2)
e (1)
e (1)
(a)
FIGURE 5.8. Stress states in the case of simple shear.
(b)
In the case where =
+ (Figure 5.8a), the stress matrix is:

0 0 0 0 . 0 0 0 (5.40)
5.3.5 State of Plane Stresses

A stress state is a state of plane stresses at a point, if and only if one of the principal stresses is zero. For example: (3) = 0 . The stress matrix is then written in the principal basis as:
(1) 0 0 (2) 0 0 0 0 . 0
(5.41)
In a basis obtained from the principal basis by rotation around the direction (3) e , it is found that the stress matrix may be written in the form:
11 12 12 22 0 0 0 0 . 0
(5.42)
As properties, we have: if n is collinear with e (3) , then t ( M , n ) = 0 ,

every stress vector is contained in the vector plane ( e (1) , e (2) ) .
5.4 Engineering Matrix Notation
79
5.3.6 Arbitrary Stress State

Given the linear properties of tensors, an arbitrary stress state at a point M can always be considered as the superposition (or more generally as a linear combination) of particular stress states. For example : an arbitrary stress state can be considered as the superposition of three uniaxial tensions (or compressions) in the principal directions; a shear stress state can be considered as the superposition of a tension and a compression of the same magnitudes (Subsection 5.3.4).
5.4 ENGINEERING MATRIX NOTATION 5.4.1 Introduction of the Notation

The symmetry of the stress tensor reduces to six the number of components necessary for defining this tensor. For this reason, engineers use a notation which consists in replacing the tensor table of nine components by a column matrix with six components, using one of the notations:
1 11 xx 2 22 yy 3 33 zz . = = yz 4 23 5 13 xz 6 12 xy
(5.43)
This notation allows to condense the forms of the elasticity relations which will be considered in Chapter 7. However, the determination of the stress vector requires to revert to the 3 3 matrix notation.

With this new notation, the relations between the stresses i expressed in the ' basis ( e1 , e2 , e3 ) and the stresses i expressed in the basis ( e1' , e2 , e3 ) , obtained by a rotation around the direction e3 , can be written in a matrix form deduced
80
Chapter 5 Stresses
from Expressions (5.20), as:

2 cos 1 2 2 sin 3 0 = 0 4 5 0 sin cos 6
sin 2 cos 2 0 0 0 sin cos
0 0
0 0
0 0 0 sin cos 0
1 0 0 cos 0 sin 0 0
2sin cos 1 2sin cos 2 0 3 . (5.44) 4 0 0 5 cos 2 sin 2 6
Thus, in the condensed matrix form, the transformation relation of stresses is written as:
= T ,
(5.45)
where T is the transformation matrix of the stresses:

cos 2 sin 2 0 T = 0 0 sin cos sin 2 cos 2 0 0 0 sin cos 0 0 0 0 0 sin cos 0 2sin cos 2sin cos 0 . (5.46) 0 0 2 2 cos sin
0 0 1 0 0 cos 0 sin 0 0
The inverse relation is obtained by inverting Equation (5.45). Thus, we have:

1 = T .
(5.47)
1 is in fact easily deduced from the transformation matrix The inverse matrix T T by changing into , since the inverse transformation corresponds to a rotation of around the direction e3 . Whence:
cos 2 sin 2 0 1 T = 0 0 sin cos
sin 2 cos 2 0 0 0 sin cos
0 0
0 0
0 0 0 sin cos 0
1 0 0 cos 0 sin 0 0
2sin cos 2sin cos 0 . (5.48) 0 0 2 2 cos sin
Exercises
81
EXERCICES
5.1 The stress state at a point M of a structure is defined in the reference system (Mxyz) by:
30 10 0 (M ) = 10 10 0 MPa . 0 0 5
Determine the values of the principal stresses at M and the principal direction of stresses. Express the stress vector t ( M , n ) which is exerted at the point M on a unit surface with orientation of the normal n that has components (nx , n y , nz ) with respect to the reference system (Mxyz). Study the cases where the unit surface is orthogonal to Mx ; orthogonal to My ; orthogonal to Mz . Derive the maximum value of the tangential stress and the corresponding direction of the surfaces.
5.2 Express the transformation matrices T in the case where the direction e3 is reversed: ( e1 , e2 , e3 ) ( e1 , e2 , e3 ) ; and then in the case where the direction e2 is reversed: ( e1 , e2 , e3 ) ( e1 , e2 , e3 ) .
5.3 In the case of a state of plane stresses in the plane (1, 2), derive the 3 3 transformation matrix for a rotation around the direction 3.
CHAPTER 6
Strains
6.1 STRAIN STATE AND STRAIN TENSOR 6.1.1 Strain State at a Point
We consider a deformable solid (S) (Figure 6.1). In its non-deformed state, the position of an arbitrary point M of the solid (S) is defined, in a coordinate system ( O / e1, e2 , e3 ) , by the position vector:
OM = x1e1 + x2e2 + x3e3 .
(6.1)
Under the effect of external mechanical loading, the solid deforms and its points are displaced: the point M moves to the point M . The position of the point M in the deformed state is defined by: OM = OM + MM . (6.2) The displacement vector MM is a function of the point M, and is generally denoted as u ( M ) . This displacement varies continuously in the interior of the solid. For another point N of the solid (Figure 6.1), we have similarly: ON = ON + NN avec NN = u ( N ) , (6.3)
where N and N are the respective positions in the non-deformed and deformed states. The relative position of the points N and M, after deformation, is given by: M N = ON OM = ON OM + u ( N ) u ( M ) , which yields: M N = MN + u ( N ) u ( M ) .
(6.4)
6.1 Strain State and Strain Tensor
83
N M M
(S)
after deformation before deformation

FIGURE 6.1. Solid deformation.
If the points M and N are infinitely close, we have: MN = ON OM = dOM . The vector dOM represents the variation of the position vector OM , when one goes from the point M to the point N. Similarly: M N = ON OM = dOM . Expression (6.4) is then written as: dOM = dOM + du ( M ) . (6.5) The scalar magnitudes dOM and dOM represent the distances between the infinitesimally close points M and N, respectively in the non-deformed state and the deformed state. We can thus write:
dOM
2
2 2 2 = ( dOM ) = ( dOM ) + 2dOM du ( M ) + ( du ( M ) ) .
The variation between the square of the distances between the points M and N from the non-deformed state to the deformed state is thus expressed as:
2 2 2 dOM dOM = 2dOM du ( M ) + ( du ( M ) ) . The vector d OM is expressed in the basis ( e1 , e2 , e3 ) by the relation: dOM = e1 dx1 + e2 dx2 + e3 dx3 ,
(6.6)
or
d OM = ei dxi ,
i = 1, 2, 3,
on introducing the summation convention in which a repetition of a subscript indicates a summation.
84
Chapter 6 Strains
The displacement vector u ( M ) is similarly written as: u (M ) = e ju j , where uj are the components f the vector u ( M ) . Whence:
u du ( M ) = ek k dx j x j j , k = 1, 2, 3 ,
(6.7)
With a double summation over the subscripts j and k. Expression (6.6) may thus be written as:
( dOM )2 ( dOM )2 = 2 ui

x j
dxi dx j +
uk uk dxi dx j . x j xi
Permutation of subscripts i and j does not change the value of the sum, since:
u j ui dxi dx j = dx j dxi . x j xi Whence the preceding relation can be put in the form:
( dOM )2 ( dOM )2 = ui

u j uk uk + + dx dx . x j xi xi x j i j
(6.8)
6.1.2 Strain Tensor

Relation (6.8) can be expressed as:
( dOM )2 ( dOM )2 = 2 ij dxi dx j ,

By introducing :
i ij = + + k k . 2 x j xi xi x j

(6.9)
1 u
u j
u u
(6.10)
The nine quantities ij constitute the components of the strain tensor in the basis ( e1, e2 , e3 ) . For a linearized theory, called theory of small deformations, the
uk uk is neglected in comparison with those of xi x j the first-order. The components of the strain tensor are then reduced to:
second-order infinitesimal term
i + ij = . x 2 j xi
1 u
u j
(6.11)
So, the strain tensor is a symmetric second-order tensor:
ji = ij .
(6.12)
85
It is written in the matrix form as:

11 12 13 (M ) = 12 22 23 , 13 23 33
(6.13)
with
11 =
1 u u 1 u u u1 , 12 = 1 + 2 , 13 = 1 + 3 , 2 x2 x1 2 x3 x1 x1
22 =
u2 , x2
23 = 2 + 3 , 2 x3 x2 33 =
u3 . x3
1 u
(6.14)
6.1.3
Interpretation of the Strain Components
We consider at the point M (Figure 6.2) two infinitesimal elements MN1 and MN2, in the non-deformed state, respectively parallel to the axes Ox1 and Ox2 : MN1 = e1 dx1, MN 2 = e2 dx2 .
and N 2 . After deformation, the points M, N1 and N2 are respectively at M , N1 We have: = u ( M ) + d1u ( M ) , M N1
with, from (6.7):
u u d1u ( M ) = e1 1 dx1 + e2 2 dx1 . x1 x1

Thus, the quantity
(6.15)
u1 dx1 = 11 dx1 represents the extension, or elongation, of the x1 element MN1 in the direction e1 . From this it results that 11 represents the elongation per unit length, called the normal strain or simply the strain, at point M in the direction e1 . Similarly:
= u ( M ) + d 2u ( M ) , M N 2 u u d 2u ( M ) = e1 1 dx2 + e2 2 dx2 . x2 x2
with: (6.16)
86
Chapter 6 Strains
x2
u2 dx x2 2 u (M )
N2
u1 dx x2 2
d 2u ( M )
N2
d1u ( M )
N1
1
u (M )
M N1
u2 dx x2 1
u (M )
u1 dx x1 1
x1
Figure 6.2. Deformations at a point.
The quantity
u2 = 22 thus represents the elongation per unit length, or the x2 normal strain, at point M in the direction e2 . before and after deforThe angle 1 between the directions MN1 and M N1 mation is such that: u2 u2 dx1 u x1 x1 = 2, tan 1 = u u dx1 + 1 dx1 1 + 1 x1 x1 x1
u Since the deformations are small 1 1 . Finally, we can write: x1
is: Similarly, the angle 2 between the directions MN 2 and M N 2
u2 . x1
u1 . x2
87
It results that the angle, originally equal to /2 between the directions MN1 and MN 2 at point M before deformation, is decreased after deformation by the angle 12 = 1 + 2 , with:
12 = 212 .
(6.17)
The component 12 of the strain tensor thus represents half the angular deformation at point M between the orthogonal directions MN1 and MN 2 . This angular deformation 12 is called the shear strain at point M between the directions e1 and e2 . Similarly : the component 33 =
direction e3 ;
u3 represents the normal strain at point M in the x3
1 u u the component 13 = 1 + 3 represents half the shear strain 13 2 x3 x1 between the directions e1 and e3 ; 1 u u the component 23 = 2 + 3 represents half the shear strain 23 2 x3 x2 between the directions e2 and e3 .
6.1.4 Compatibility Conditions

The displacement field corresponding to a given deformation state is obtained by integrating Equations (6.14). It results that there is a total of six equations to derive the three components (u1, u2, u3) of the displacement. In order that the solution be unique, the components of the strain tensor must satisfy the six relations:
2 ii 2 x2 j + 2 ij 2 xi2 =2 2 ij xi x j , i j, , i j k,
2 ii jk ik ij = + + x j xk xi xk x x i j with i, j , k = 1, 2, 3.
(6.18)
These expressions constitute the six compatibility relations.
88
Chapter 6 Strains
6.2 STRAIN STATE AT A POINT 6.2.1 Elongation per Unit Length

We consider an arbitrary element MN of a solid, of length dl and the direction of which is defined by the unit vector u of the bipoint (M, N). We have: MN = u dl . The components (1, 2, 3) of the vector u in the basis ( e ) = ( e1, e2 , e3 ) are the direction cosines defining the direction of the segment MN. The components of the vector MN in the basis ( e ) are:
dx = 1 dl , 1 MN dx2 = 2 dl , d x = dl . 3 3
(6.19)
We look for the elongation per unit length of MN, that is the normal strain in the direction (1, 2, 3). After deformation, we have:
M N = dl + dl .
So, Expression (6.9) is written as:

(1 + )2 1 ( dl )2 = 2 ij i j ( dl )2 .
2 Whence on neglecting , it yields :
= i j ij .
When expanded this relation becomes:
2 2 = 1211 + 2 22 + 3 33 + 21 212 + 21 313 + 2 2 3 23 .
(6.20)
(6.21)
This expression has the same form as Expression (5.15) which expresses the normal component tn of the stress. Thus, it is deduced that the normal strain in the direction u can be expressed in forms similar to Expressions (5.13) and (5.14). Thus: = u [ ( M ) u ] , (6.22) Or in matrix form:
t = u ( M ) u .
(6.23)
6.2.2 Shear Strain

The shear strain allows to express the variation of the angle between
6.2 Strain State at a point
89
the two directions u and u , in the non-deformed state, which become u and u , after deformation (Figure 6.3). After deformation, the angle between the two directions u and u is
+ d , where d is the shear strain. We can show that in matrix form we

have:
t t t d ( cos ) = d u u = 2u ( M ) u ( + ) u u ,
(6.24)
an expression which allows to derive d . The expansion of this expression leads to: d ( cos ) = 2 (1111 + 2 2 22 + 3 3 33 ) + 2 (1 2 + 2 1 ) 12
+ 2 ( 2 3 + 3 2 ) 23 + 2 ( 3 1 + 1 3 ) 13 ( + ) (11 + 2 2 + 3 3 ) ,
(6.25)
where (1, 2, 3) and (1, 2, 3) are the components of the directions u and u , and and are the extensions per unit length in the same directions.
Relation (6.24) can be rewritten as:

t sin d = 2u ( M ) u ( + ) cos ,
since:
t u u = cos .
Finally, the shear strain can be expressed as:

d = 1 2 t + ) u ( M ) u . ( tan sin
(6.26)
In the case where the two directions and are orthogonal, we have: t u u = u u = cos = 0 . u
u
u
u
M FIGURE 6.3. Shear deformation.
90
Chapter 6 Strains
Then denoting the two directions u and u by u1 and u2 , Expression (6.24) is written as:
t t d ( cos 12 ) = d ( u1 u 2 ) = 2u1 ( M ) u 2 , t t d ( cos 12 ) = d ( u 2 u1 ) = 2u 2 ( M ) u1 .
or
(6.27)
The shear strain is then given by:

t t d12 = 2u1 ( M ) u 2 = 2u 2 ( M ) u1 .
(6.28)
6.2.3
Strain Tensor in the Principal Directions

(1) 0 ( M ) = 0 (2) 0 0 0 0 , (3)
Referred to its principal directions, the strain tensor is:
(6.29)
where (1) , (2) and (3) are the principal strains . The elongation per unit length in the direction (1, 2, 3), referred to the principal directions ( e (1) , e (2) , e (3) ) is expressed, from (6.21), as:
2 (1) 2 (2) 2 (3) = 1 + 2 + 3 .
(6.30)
The elongations per unit length in the three principal directions are:
1 = (1) ,
2 = (2) ,
3 = (3) .
(6.31)
The shear strain is, by (6.25), then written as:

d ( cos ) = 2 (11 (1) + 2 2 (2) + 3 3 (3) ) ( + ) (11 + 2 2 + 3 3 ) .
(6.32)
If the two directions and are orthogonal, the preceding expression simplifies as: d ( cos ) = 2 (11 (1) + 2 2 (2) + 3 3 (3) ) . (6.33)
If, furthermore, these two directions coincide with the principal directions, we have: d ( cos ) = 0 . So, in this case there is no shear deformation. Lastly, in the case where the direction u is the same as a principal direction (for example u = u (3) ), Relation (6.24) can be rewritten by taking account of: ( M ) u = ( M ) u (3) = (3)u(3) .
6.2 Strain State at a point
91
So, then:
t (3) d ( cos (3) ) = ( (3) ) u u .
(6.34)
In particular, if the direction u is orthogonal to the principal direction u (3) , we

t (3) u = 0 , and thus d ( cos (3) ) = 0 . So, in the plane orthogonal to an have u
eigendirection, all the directions vary under deformation while staying in this plane.

Since the strain tensor has a table of components similar to that of the stress tensor, the transformation relations are expressed in the same form as for the stress tensor (Section 5.2.4). For a rotation around the direction e3 , the relations are transposed from Expressions (5.20), whence:
= 11 cos 2 + 212 sin cos + 22 sin 2 , 11 = ( 22 11 ) sin cos + 12 (cos 2 sin 2 ), 12 = 13 cos + 23 sin , 13 = 12 , 21 = 11 sin 2 212 sin cos + 22 cos 2 , 22 = 13 sin + 23 cos , 23 = 13 , 31 = 23 , 32 = 33 . 33
(6.35)
In the case where the initial basis is the principal basis, the preceding relations are reduced to:
= (1) cos 2 + (2) sin 2 , 11 = ( (2) (1) ) sin cos , 12 = 0, 13 = 12 , 21 = (1) sin 2 + (2) cos 2 , 22 = 0, 23 = (3) . 33 = 0, 31 = 0, 32
(6.36)
92
Chapter 6 Strains
6.3 PARTICULAR DEFORMATION STATES 6.3.1 Spherical Tensor and Strain Deviator
Like in the case of the stress tensor, it is possible to decompose the strain tensor into a spherical tensor and a strain deviator, as:
( M ) = S ( M ) + D ( M ) ,
(6.37)
with
e S ( M ) = I and e = tr ( M ) . 3
(6.38)
The tensor D(M) is a tensor with zero trace, having the same principal directions as (M) and for principal strains:
(1) ,
e 3
(2) ,
e 3
(3) .
e 3
(6.39)
The deformation resulting from the deviator D(M) is of zero cubic dilatation: that is the deformation occurs without a change of volume.
6.3.2 Particular States

1. A deformation state at a point is a state of uniform dilatation, if and only if the deviator is zero. The elongations per unit length are equal whatever the direction. 2. A deformation state is uniaxial (simple extension), if and only if two of the principal deformations are zero. 3. A deformation state at a point is a state of simple shear, when one of the principal strains is zero and the other two are opposites. 4. A deformation state at a point is a state of plane deformations, if and only if one of the principal deformations is zero. 5. An arbitrary deformation state can always be considered as the superposition of particular deformation states.
6.4 ENGINEERING MATRIX NOTATION 6.4.1 Introduction of the Notation

As for the stress tensor, the symmetry of the strain tensor reduces to six the number of components necessary for determining the strain tensor. For this reason, engineers use a notation that consists in replacing the matrix table of nine
6.4 Engineering Matrix Notation
93
components of the tensor by a column matrix of six components, and in substituting for the components 12, 13, 23 the shear deformations 12, 13, 23. Thus, engineers use one of the following notations:
1 11 11 xx xx 2 22 22 yy yy 3 33 33 zz zz . = = = = 2 2 yz yz 4 23 23 5 13 213 xz 2 xz 2 6 12 212 xy xy
(6.40)
6.4.2 Change of the Reference System

With this new notation, the relations between the strains i expressed in the ' basis ( e1 , e2 , e3 ) and the deformations i expressed in the basis ( e1' , e2 , e3 ) , obtained by rotation around the direction e3 , can be written in a matrix form deduced from Expressions (6.35), as:
2 cos 1 2 2 sin 3 0 = 0 4 5 0 6 2sin cos
sin 2 cos 2 0 0 0 2sin cos
0 0
0 0
0 0 0 sin cos 0
1 0 0 cos 0 sin 0 0
1 sin cos 2 0 3. 4 0 0 5 cos 2 sin 2 6 sin cos (6.41)
Whence, in condensed form:

= T ,
(6.42)
where T is the transformation matrix for strains: cos 2 sin 2 0 T = 0 0 2sin cos sin 2 cos 2 0 0 0 2sin cos 0 0 0 0 0 0 0 sin cos 0 sin cos 0 . (6.43) 0 0 2 2 cos sin sin cos
1 0 0 cos 0 sin 0 0
94
Chapter 6 Strains
The inverse transformation relation is obtained by inverting Expression (6.42), which yields:
= T1 ,
(6.44)
where the inverse matrix T1 is deduced from the matrix T by changing into : cos 2 sin 2 0 0 0 sin cos sin 2 cos 2 0 0 0 sin cos 0 0 1 0 0 0 . (6.45) T-1 = 0 0 0 cos sin 0 0 0 0 sin cos 0 2 2 0 0 cos sin 2sin cos 2sin cos 0 The difference between these expressions for the strain transformation matrix and the ones for the stress transformation matrix results from the introduction of the factor two, when the subscripts for the strains 4, 5 and 6 are contracted. As a function of the results obtained, it can be written:
1 ) T = ( T t
and
t T1 = T .
(6.46)
These relations are not general. They are satisfied only in the case of a rotation around the direction 3 of the initial basis.
EXERCISES
6.1 We consider a plate of surface (S), on which is drawn a square meshing of elementary side l (Figure 6.4). This plate is subjected to a deformation state such that: 1. The direction orthogonal to the surface (S) remains orthogonal to the surface during the deformation. The unit strain through the thickness is 3.
B
B l C
2
M
A
A
FIGURE 6.4. Deformation of square meshing.
Exercises
95
2. After deformation, the elementary mesh MACB at point M is deformed to M AC B such as:
the directions MB and MA make an angle of + . 2 6.1.1 Determine the strain tensor at point M in the case where:
l = 10 mm,
MA = MA MA = l1 , MB = MB MB = l2 ,
l1 = 3 m,
l2 = 1 m,
= 5 104 rad,
3 = 3 104.
6.1.2 Show that the direction orthogonal to the plate is a principal direction. Derive the principal strains and the principal directions. 6.1.3 In the plane of the plate, determine: the directions of zero elongation; the directions for which he strain is extremal; the directions for which there is no angular deformation. et My1 of the plane of the plate, 6.1.4 Now, we consider the directions Mx1 obtained by a rotation of angle = 20 from the directions MA and MB . ; the ; the stain in the direction My1 Determine: the strain in the direction Mx1 and My1 . shear strain of the two directions Mx1 6.1.5 By considering the transformation relations, determine the strain tensor , My1 , direction orthogonal to the components in the reference system ( Mx1 plate). Compare the results with the ones obtained in 6.1.4.
6.2 In the case of a state of plain strains in the plane (1, 2), derive the 3 3 transformation matrix, in the case of a rotation of angle around the direction 3. 6.3 We consider a deformable solid subjected to a deformation state. At point M on the surface of the solid, the tangent plane (P) can be considered as being identified with the surface. This plane is the principal plane of the strain tensor at point M. In order to determine the strain tensor, we put at point M a rosette of three strain gauges of which the directions ( Mx1, Mx2 , Mx3 ) make angles 12, 23, 13 with each other (Figure 6.5). The electrical resistance of a strain gauge varies as a function of the gauge elongation, and measurement of this variation allows to deduce the corresponding strain. When the solid is deformed, the strains measured are respectively 1, 2, 3 in the directions ( Mx1, Mx2 , Mx3 ) . The object of the problem is to deduce: the
eigenvalues
(1)
and
(2)
in the principal plane (P), the orientation of the eigen-
96
Chapter 6 Strains
direction (1) with the direction Mx1 .
Hereafter, we shall note as , the angle (Figure 6.5) between the direction Mx1 and the eigendirection (1).
6.3.1 Derive the relations that express 1, 2 and 3 as functions of sin , cos , and the principal strains , . 6.3.2 The three strain gauges are usually arranged at 45 (12 = 13 = /4). In this case, derive the relations which express 1, 2, 3 as functions of , sin and cos .
(1) (1) (2)
(2),
From these, deduce the expression which allows to obtain, as function of 1, 2 and 3, the angle between the direction Mx1 and the eigendirection (1). Application:
1 = 4 104 ,
2 = 5 104 ,
3 = 2 104.
(1)
Next, find the expressions for the eigenvalues application. x3
and . Do the numerical
(2)
13
x2
23 12
x1
(1)
FIGURE 6.5. Strain gauge arrangement.
CHAPTER 7
Elastic Behaviour of Materials
7.1 STRESS-STRAIN RELATIONS FOR ANISOTROPIC MATERIALS 7.1.1 Introduction

The strain and stress fields in a material are related by laws of behaviour which characterize the mechanical behaviour of the material. These laws are deduced from experimental investigations and allow us to have the best description of the experimental processes observed. Experience shows that most materials have, at a given temperature, a linear elastic behaviour. This behaviour is considered hereafter in this chapter.
7.1.2 Stiffness Matrix

The linear elasticity relation can be written in the following matrix form: 1 C11 C 2 12 3 C13 = 4 C14 5 C15 6 C16 or in the condensed form:
= C .
C12 C22 C23 C24 C25 C26
C13 C23 C33 C34 C35 C36
C14 C24 C34 C44 C45 C46
C15 C25 C35 C45 C55 C56
C16 1 C26 2 C36 3 , C46 4 C56 5 C66 6
(7.1)
(7.2)
This law, usually called the generalized Hookes law, introduces the symmetric stiffness matrix C. So, the linear behaviour of a material is described in the general case by 21 independent coefficients, here the 21 stiffness constants Cij .
98
Chapter 7 Elastic Behaviour of Materials
7.1.3 Compliance Matrix

The elasticity relation (7.2) can be written in the inverse form as:
= S ,
(7.3)
by introducing the inverse of the stiffness matrix. The matrix S is called the compliance matrix, and in the general case is written as:
S11 S 12 S13 S= S14 S15 S16
S12 S22 S23 S24 S25 S26
S13 S23 S33 S34 S35 S36
S14 S24 S34 S44 S45 S46
S15 S25 S35 S45 S55 S56
S16 S26 S36 , S46 S56 S66
(7.4)
with
S = C1 .
(7.5)
The coefficients Sij are called the compliance constants.
7.1.4 Change of the Reference System

Let C = Cij be the stiffness matrix expressed in the basis ( e ) = ( e1, e2 , e3 ) , e2 , e3 ). and C = Cij the stiffness matrix expressed in the basis ( e ) = ( e1 These matrices relate the stresses and strains expressed in the respective bases: =C, (7.6) in the basis ( e ) : = C . (7.7) in the basis ( e ) : The transformation relations for stresses and strains can be written in the general case in the forms (5.45)(5.47) and (6.42)(6.44). Combining Relations (5.47) and (7.6) yields:
= T = T C ,
then, taking account of (6.44), we obtain:

= T C T1 .
(7.8)
The identification of Expressions (7.7) and (7.8) then leads to the transformation relation of the stiffness matrices:
C = T C T1 .
(7.9)
7.1 Stress-Strain Relations for Anisotropic Materials
99
In a similar way, we find the transformation relation for the compliance matrices as:
1 S = T S T .
(7.10)
The inverse relations are respectively written as:

1 C = T C T ,
(7.11) (7.12)
and
S = T1 S T .
7.1.5 The Different Types of Anisotropic Materials

7.1.5.1 Introduction
In the most general case, the stiffness matrix and the compliance matrix are each determined by 21 independent constants. This case corresponds to a material which haves no symmetry property. Such a material is called a triclinic material. Most anisotropic materials have a structure which exhibits one or more symmetries. For example, it is the case of monocrystals, fibrous structures, fibre or cloth composite materials, etc. The properties of the geometric symmetries then reduce the number of independent constants needed to describe the elasticity behaviour of the materials. This reduction is a function of the symmetries exhibited by the material considered.
7.1.5.2 Monoclinic Material

A monoclinic material is a material which has a plane of symmetry. The form of the stiffness (or compliance) matrix must be such that a change of the reference system obtained by a symmetry about this plane does not modify the matrix. In the case where the plane of symmetry is the (1, 2) plane, the consideration of the transformation relations leads to a stiffness matrix of the form: 0 C11 C12 C13 C 0 12 C22 C23 C13 C23 C33 0 0 0 C44 0 0 0 0 C45 0 C16 C26 C36 0 C16 0 C26 0 C36 . C45 0 C55 0 0 C66
(7.13)
The compliance matrix has the same form. Thus, the number of independent elasticity constants is reduced to 13.
100
7.1.5.3 Orthotropic Material

An orthotropic material has tree symmetry planes that are mutually orthogonal. In fact, the existence of two orthogonal symmetry planes implies the existence of the third. Thus, the form of the stiffness matrix is deduced from the preceding one of the monoclinic material by adding a second symmetry plane orthogonal to the first one. The invariance of the stiffness matrix under a change of reference system obtained by symmetry about this second plane leads to a stiffness matrix of the form: 0 C11 C12 C13 C 0 12 C22 C23 C13 C23 C33 0 0 0 C44 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 . 0 0 0 C55 0 C66
(7.14)
The compliance matrix has the same form. Thus, the number of independent elasticity constants is reduced to 9.
7.1.5.4 Unidirectional Material

The elementary cell of a unidirectional composite material can be considered as being constituted of a fibre embedded in a cylinder of matrix (Figure 7.1). The material thus behaves as an orthotropic material having one axis of revolution in addition. The material is then called as a revolution orthotropic or transverse isotropic material. It results that a change of the reference system obtained by an arbitrary rotation around the revolution axis must leave unchanged the stiffness (or compliance) matrix. The application of this property leads to:
C13 = C12 , C55 = C66 , C33 = C22 , C44 =
1 2
( C22 C23 ) .
(7.15)
matrix fibre 1
3
FIGURE 7.1. Unidirectional composite material.
7.2 Isotropic Materials
101
and S13 = S12 , S55 = S66 , S33 = S22 , S44 = 2 ( S22 S23 ) . (7.16)
The stiffness matrix is therefore written as:

C11 C12 C12 C 12 C22 C23 C12 C23 C22 0 0 0 0 0 0 0 0 0 And the compliance matrix is: S11 S 12 S12 0 0 0 S12 S22 S23 0 0 0 S12 S23 S22 0 0 0 0 0 0 2 ( S22 S23 ) 0 0 0 0 0 0 S66 0 0 0 0 . 0 0 S66 0 0 0 1 C C 23 ) 2 ( 22 0 0 0 0 0 0 C66 0 , 0 C66 0 0 0 0
(7.17)
(7.18)
The elasticity properties of a unidirectional material are described by 5 independent elasticity constants.
7.2 ISOTROPIC MATERIALS 7.2.1 Stress-Strain Relations

A material is isotropic if its properties are independent of the choice of the reference system. The usual materials (with the exception of the wood) satisfy this description established at the macroscopic scale. There then exists no privileged direction and the stiffness and compliance matrices must be invariant under every change of orthonormal reference system. The application of this property to a unidirectional material leads to the relations:
C22 = C11 , C23 = C12 , C66 =
1 2
( C11 C22 ) .
(7.19)
The number of independent elasticity constants is thus reduced to 2 and the stiffness matrix is written as:
102
C11 C12 C12 C 12 C11 C12 C12 C12 C11 0 0 0 0 0 0 0 0 0
0 0 0
1 2
0 0 0 0
1 2
( C11 C12 )
0 0
( C11 C12 )
0
0 0 . 0 0 1 C C 12 ) 2 ( 11 0
(7.20)
Usually, the stiffness constants are expressed by introducing the Lam coefficients and :
C12 = ,
1 2
( C11 C22 ) = ,
(7.21) (7.22)
hence
C11 = + 2 .
The elasticity relation (7.1) can thus be written as:
ij = ij kk + 2 ij ,
or
(7.23) (7.24)
ij = ij tr + 2 ij ,
where tr = kk = 11 + 22 + 33 is the volumetric strain of the isotropic material. The normal stresses (j = i) are therefore written as:
ii = tr + 2 ii ,
And the tangential stresses (j i):
(7.25)
ij = 2 ij .
From Relation (7.25), we deduce:
(7.26)
11 + 22 + 33 = ( 3 + 2 )( 11 + 22 + 33 ) ,
or
tr = ( 3 + 2 ) tr .
(7.27)
The strains as functions of the stresses are easily derived from Relations (7.25) to (7.27). The normal strains are written as:
ii =
and the tangential strains are:
1 ii , tr + 2 ( 3 + 2 ) 2
1 ij . 2
(7.28)
ij =
(7.29)
103
The two relations (7.28) and (7.29) can be summarized by the expression:
ij =
1 ij tr + ij . 2 ( 3 + 2 ) 2
(7.30)
7.2.2 Moduli of Elasticity

The elasticity relations are usually expressed as functions of the engineering constants. For an isotropic material, these constants are the Youngs modulus, the Poissons ratio and the shear modulus, as well as some other constants that will be discussed in the present section. These engineering constants are measured in simple mechanical tests in which the material is subjected to a particular state of stresses and strains.
7.2.2.2 Uniaxial Tension or Compression

In the case of a uniaxial tension or compression in the 1-direction, the stress matrix is (Section 5.3.3): 1 0 0 0 0 0 . 0 0 0 It results that the tangential strains (ij , with j i) are zero. The normal strain in the test direction is, by (7.28):
1 = 11 =
Thus:
1 1 + 1 . 2 ( 3 + 2 ) 2 1 = E1 ,
(7.31) (7.32)
on setting:
E=
( 3 + 2 ) . +
Relation (7.31) is the Hookes law, and the coefficient E is the Youngs modulus. The transverse strains are given similarly by (7.28) as:
2 = 3 =
or
1 = 1 , 2 ( 3 + 2 ) 2 ( + )
(7.33)
2 = 3 = 1 = 1 ,
E
104
on setting:
. 2( + )
(7.34)
The parameter is the Poisson ratio of the material.
7.2.2.3 Simple Shear Test

In a shear test, the stress matrix referred to the directions (1, 2) making an
4 Therefore:
angle of
+ with respect to the stress principal directions is given by (5.40).
11 = 22 = 33 = 0,
13 = 23 = 0,
12 = .
(7.35)
The stress-strain relations are then reduced to:
12 = 212 = 12 ,
(7.36)
where 12 is the shear strain between the two orthogonal directions 1 and 2. The coefficient appears as being the shear modulus. This modulus is more usually denoted by G. The other strains are zero.
7.2.2.4 Spherical Compression or Tension

In such a test, the stresses are spherical. It results that:
ij = p ij .
(7.37)
In practice, it is used a hydrostatic compression: p < 0. The principal stresses are equal:
11 = 22 = 33 = 1 tr = p , 3
(7.38)
and the tangential stresses are zero: 12 = 13 = 23 = 0 . It results that the strains are also spherical:
11 = 22 = 33 = 1 tr , 3
with
(7.39) (7.40)
p= +2 tr . 3
The coefficient:
k =+2 3
is the bulk modulus.
(7.41)
105
7.2.3 Relations between the Elasticity Coefficients

The elastic properties of an isotropic material are characterized by two independent coefficients. Table 7.1 summarizes the relations between the different parameters as functions of the pairs (, ), (E, ) and (E, G). The coefficients E and are usually considered, because they are derived easily in a traction test. The elasticity relation (7.30) is then written as:
ij =
ij tr +
1 + ij . E
(7.42)
7.2.4
Stiffness and Compliance Matrices
As functions of the various elasticity coefficients introduced, the stiffness and compliance matrices can be written in the form:
a b b 0 0 0
b a b 0 0 0
b b a 0 0 0
0 0 0 c 0 0
0 0 0 0 c 0
0 0 0 , 0 0 c
(7.43)
TABLE 7.1. Relations between the elasticity coefficients of an isotropic material.
E,
E (1 + )(1 2 ) E 2 (1 + )
E, G G ( E 2G ) 3G E G
( 3 + 2 ) + 2 ( + )
E E 2G 2G
GE 3 ( 3G E )
E 3 (1 2 )
+2 3
106
with, for the stiffness matrix C:

a = + 2 = b= = c== E (1 ) 4G E =G , 3G E (1 + )(1 2 )
E E 2G =G , 3G E (1 + )(1 2 ) E = G, 2 (1 + )
(7.44)
and for the compliance matrix S:

a=
+ 1 = , ( 3 + 2 ) E = , 2 ( 3 + 2 ) E
= 2 (1 + ) E = 1 . G
b= c= 1
(7.45)
EXERCISES
7.1 In the case of a monoclinic material with the symmetry plane (1, 2), show that the stiffness matrix has the form (7.13). 7.2 The symmetry plane (1, 3) is added to a monoclinic material in order to obtain an orthotropic material. Show that the stiffness matrix is of the form (7.14). 7.3 We consider a rotation of angle around the 1-direction. Express the stiffness matrix in the new directions. 7.4 From the preceding exercise, deduce the form (7.17) of the stiffness matrix of a transverse isotropic material. 7.5 We consider an orthotropic material of directions (1, 2, 3), and at a point of this material, it is applied a deformation state with principal directions (1', 2', 3) = 5 103 , 2 = 2 103 , 3 = 4 103 . The directions (1', and principal strains 1 2') make an angle = 30 with the directions (1, 2) of the material. The stiffness constants of the material in directions (1, 3) are:
Exercises
107
C11 = 32 GPa, C22 = 20 GPa, C44 = 2.5 GPa,
C12 = 4.2 GPa, C23 = 4 GPa, C55 = 3.5 GPa,
C13 = 3.8 GPa, C33 = 12 GPa, C66 = 4 GPa.
Determine the stress matrix in the directions (1, 2, 3), and then the principal strains and directions.
CHAPTER 8
The Mechanics of Deformable Solids
8.1 FUNDAMENTAL RELATION FOR DEFORMABLE SOLIDS

Let (D) be a domain of a continuous medium, with boundary (S) (Figure 8.1). The forces applied to the domain (D) are of two types: 1. The body forces (for example gravitational, inertia, magnetic forces, etc.) which are applied to the entire volume, characterized by the force density f ( M , t ) per unit volume. The force which acts on a volume element d V surrounding point M is: d f (M , t ) = f (M , t ) d V . (8.1)
2. The surface forces which are only applied on the boundary (S) ofthe domain (D). They are characterized by the force density (stress vector) t ( M , t ) per unit surface. The force which acts on the surface element d S surrounding point M of (S) is: d t (M , t ) = t (M , t ) d S . (8.2) The orientation of the stress vector is taken, as in Chapter 5, to be positive from the interior to the exterior of the element, and corresponds in this case to a tension (or traction) at point M. In the opposite case, a compression is exerted at M. In the case where the coordinate system (T) = (O / 123) is a Galilean reference system, the fundamental relation of the Dynamics applied to the element surrounding the point M is written as: ( M , t ) a (T ) ( M , t ) d V = f ( M , t ) d V + t ( M , t ) d S , (8.3)
where ( M , t ) is the density at M at the instant t ; a (T ) ( M , t ) is the acceleration vector of point M, with respect to the system (T), at the instant t.
8.1 Fundamental Relation for Deformable Solids
109
3 (D) M
dV
(S) O 1
FIGURE 8.1. Domain of a continuous medium.
Extended to the whole of domain (D), the fundamental relation leads to an equation of the resultant: a f dV t d S = 0 , (8.4)
and an equation of the moment at a reference point, for example the point O: OM a f d V OM t d S = 0 , (8.5)
where the integrals are taken over the volume V of the domain and at the surface S of the boundary. Each equation leads, in a given basis, to three scalar equations. For example, by introducing the components of the different vectors, the equation of the resultant is written for the direction i as:
( fi ai ) d V +
t dS = 0,
S i S ij j
or by taking account of (5.7):
( fi ai ) d V +
n dS = 0.
ij d V . x j
(8.6)
The surface integral can be transformed into a volume integral by using the general Gauss formula, which allows us to transform a surface integral into a volume integral. Here, it leads to:
ij n j d S =
Equation (8.6) can thus be written as:
fi ai + x ij d V = 0 . j V
(8.7)
110
Chapter 8 The Mechanics of Deformable Solids
This equation must be satisfied for every element d V of the domain (D). Thus the equation reduces to: (8.8) fi ai + ij = 0 . x j The consideration of the equation of the moment leads to the same equation, called the fundamental relation. The component ai of the acceleration vector is expressed as a function of the component ui of the displacement vector as:
ai = 2ui t 2
and the fundamental relation leads to the three equations:

11 12 13 2u + + + f1 = 21 , x1 x2 x3 t 12 22 23 2u + + + f 2 = 22 , x1 x2 x3 t 13 23 33 2u + + + f3 = 23 . x1 x2 x3 t (8.9)
8.2 FORMULATION OF THE MECHANICS OF DEFORMABLE SOLIDS 8.2.1 Statement of the Problem
The general problem of the mechanics of deformable solids is posed in the following way. Let (D) be a deformable solid, bounded by a surface (S) (Figure 8.2). This solid is subjected to body forces f ( M , t ) with components fi and to conditions on the boundary (S) called boundary conditions. These conditions represented on Figure 8.2 can be of two types: 1. On one part (Su) of the boundary (S), given displacements uS ( M , t ) are prescribed: u ( M , t ) = uS ( M , t ) M ( Su ) . (8.10) No surface force is applied on part (Su). 2. On the complementary part ( S ) of (S):
( Su ) ( S ) = ( S ), ( Su ) ( S ) = ,
8.2 Formulation of the Mechanics of Deformable Solids
111
uS ( M , t )
( Su )
(S) M
( S )
tS ( M , t )
(D)
FIGURE 8.2. Boundary conditions.
given surface forces tS ( M , t ) are applied (or the absence of surface forces: tS ( M , t ) = 0 ): t ( M , t ) = tS ( M , t ) M ( S ) . (8.11)
Under the effect of the applied actions (body forces and boundary conditions), the solid (D) is deformed. The problem of the mechanics of deformable solids consists of seeking the displacement field u ( M , t ) and the stress field ( M , t ) at every point M of the solid (D), satisfying: 1. The fundamental relation (8.8): 1.1 for a problem of dynamics:
2u ij + fi = 2i , x j t i, j = 1, 2, 3,
(8.12)
1.2 for a problem of statics:

ij + fi = 0, x j 1 u u j i, j = 1, 2, 3.
(8.13)
2. The strain-displacement relations (6.11):
ij = i + , 2 x j xi
in the case of small strains.
i, j = 1, 2, 3,
112
3. The compatibility conditions (6.18):
2 ij xk xl
2 jl 2 kl 2 ik = 0, xi x j x j xl xi xk i, j , k , l = 1, 2, 3.
(8.14)
4. The boundary conditions (8.10) and (8.11) : u ( M , t ) = uS ( M , t ) on ( Su ), t ( M , t ) = tS ( M , t ) on ( S ). 5. The behaviour law of the material characterizing its mechanical behaviour. This law relates the stress field to the strain field:
ij = f ( kl ),
i, j , k , l = 1, 2, 3.
(8.15)
In the case of linear elastic behaviour, the behaviour law is described by one of the Relations (7.1) or (7.3). The problem of the mechanics can be solved by a displacement approach or a stress approach. In the displacement approach, the solution consists in seeking the displacement field u ( M , t ) or the strain tensor field ( M , t ) , guided by the conditions imposed on the boundary (Su):
u ( M , t ) = uS ( M , t )
M ( Su ).
From this, we deduce the strain tensor at every point M, and then the stress tensor which must satisfy the previous conditions 1 to 5. In the stress approach, the solution consists in seeking the stress tensor ( M , t ) , guided by the conditions imposed on the boundary ( S ) :
t ( M , t ) = ( M , t )n = tS ( M , t )
M ( S ),
(8.16)
where n is the unit vector of the outward normal at M to the surface of the solid. This tensor must also satisfy the fundamental relation. We next deduce the strain tensor and the displacement field which must verify the boundary conditions imposed on ( Su ) .
8.2.2 Equations in Cartesian Coordinates

In the preceding, we have been implicitly referred to a system of Cartesian axes ( Ox1x2 x3 ) , the points of the space being described by their Cartesian coordinates ( x1 , x2, x3 ) , and the displacement vector having the components ( u1, u2 , u3 ) . Such a notation is well adapted to tensor or matrix formulations.
113
Another notation usually used by engineers consists in considering the system of axes (Oxyz). The coordinates of points are (x, y, z) and the components of the displacement vector are denoted (u, v, w). We now write the equations in terms of this notation. 1. Stresses. The table for the stress tensor is written as:
xx xy xz (M ) = xy yy yz , xz yz zz
(8.17)
and the stress column matrix is:

1 xx 2 yy 3 zz (M ) = = . yz 4 5 xz xy 6
(8.18)
2. Strains. The table for the strain tensor has the same form as that the stress tensor and the elements of the column matrix are given by the following expressions:
1 = xx = 2 = yy = 3 = zz =
u , x v , y w , z v w , = + z y u w , + z x u v + . y x
(8.19)
4 = yz = 2 yz
5 = xz = 2 xz = 6 = xy = 2 xy =
114
3. Fundamental relation. The relation is derived from (8.9) as:

2u xx + xy + xz + f x = 2 , x y z t 2v xy + yy + yz + f y = 2 , x y z t 2w xz + yz + zz + f z = 2 . x y z t
(8.20)
4. Compatibility conditions. The six compatibility equations are deduced from Expressions (6.18):
2 xx y 2 2 yy z 2 2 zz x 2 + + + 2 yy x 2 2 zz y 2 2 xx z 2 =2 =2 2 xy xy 2 yz yz
, ,
2 xz , =2 xz
2 xx yz xz xy = + + , yz x x y z 2 yy xz = xz xy yz + + y y z x , .
(8.21)
2 zz xy yz xz = + + y xy z z x
5. Boundary conditions. The conditions on the stresses are: t ( M , t ) = tS ( M , t ) at the boundary ( S ). If (tSx , tSy , tSz ) are the components of the stress vector tS applied at point M and (nx , n y , nz ) the components of the unit vector of the normal at point M oriented from the interior to the exterior of the solid, the condition on the stresses is written as:
t x xx xy xz nx tSx t = y xy yy yz n y = tSy , tz nz tSz xz yz zz
(8.22)
115
Whence:
xx nx + xy n y + xz nz = tSx , xy nx + yy n y + yz nz = tSy , xz nx + yz n y + zz nz = tSz .

(8.23)
8.2.3 Equations in Cylindrical Coordinates

The use of cylindrical coordinates is imposed in numerous problems, in particular when the problem is one of revolution. A point M is located by its cylindrical coordinates (r, , z) (Figure 8.3). The stress and strain matrices have the forms:
rr (M ) = r rz with:
r rz z , z zz
rr and ( M ) = r rz
r rz z , z zz
w , z
(8.24)
rr =
ur u 1 u , = r + , r r r u u 1 ur 2 r = + , r r r u w 2 rz = r + , z r 1 w u + 2 z = , z r z
zz =
(8.25)
k
i
x
e j er
y r
FIGURE 8.3. Cylindrical cordinates.
116
displacement vector u in the on introducing the components (ur , u , w ) of the basis ( er , e , k ) obtained from the basis ( i , j , k ) of Cartesian coordinates, by a rotation of angle around the direction k . The relations between the components of stress matrices expressed in the bases of Cartesian coordinates and cylindrical coordinates are easily obtained considering Relation (4.13). The fundamental relations in the basis of cylindrical coordinates are next deduced by transposing Relations (8.20) in the basis ( er , e , k ) . In the case of a problem of statics, they are written as:
rr 1 r + + rr r r
rz + f r = 0, + z
r 1 + + 2 r + z + f = 0, r z r rz 1 z + + rz + zz + f z = 0. r r z
(8.26)
8.3 ENERGY FORMULATION 8.3.1 Variation of a Functional

Quite generally, a functional F can be defined as being a function of a set of functions: u(x, y, ), v (x, y, ), , and of their derivatives with respect to the set of variables (x, y, ). Whence: u u v v F = F u, , , . . . , v, , , . . . . x y x y The first variation of F is defined as:
F = u F F F u + + . . . , u + u u x u y x y
(8.27)
(8.28)
u u v v where u, , , . . . , v , , , . . . are arbitrary variations x y x y u u v v of functions u, , , . . . , v, , , ... x y x y The derivative variable u.
F is obtained by formal derivation of F with respect to the u
8.3 Energy Formulation
117
The derivative
F is obtained by formal derivation of F with respect of u x
the variable
u . It is the same for the other derivatives. x The operator has the properties of the usual variation operator d.
8.3.2 The Virtual Work Theorem

We return to the problem of mechanics (Section 8.2.1) considering, for the moment, only a problem of statics. Let u ( M ) be the displacement field that is the solution of the problem and which is written in the basis ( e1 , e2 , e3 ) as:
u ( M ) = ui ei .
(8.29)
Next, we consider the displacement field obtained by taking the first variation of u: u = ei ui , (8.30) and such that this field satisfy the conditions prescribed at the boundary (S). Thus: or u = 0, ui = 0 i = 1, 2, 3 on ( Su ), (8.31) since the displacements are prescribed on (Su). This field is called the virtual displacement field. We evaluate then the work done in this displacement, work called the virtual work. The work of the surface forces reduces, in accordance with (8.31), to:
WS =
t u d S =
S
t u d S .
S
(8.32)
Whence:
WS =
ti ui d S =
ij nij ui d S .
(8.33)
The work done by the body forces is:
WV =
f u d V =
V
fi ui d V ,
(8.34)
or, taking (8.12) into account in the case of a problem of statics, the work is:
WV =
ij
V
x j
ui d V .
(8.35)
118
By noticing that:
ui , ij ui ) = ij ui + ij ( x j x j x j
Relation (8.35) may be written as:

WV =
ij
ui d V x j
( ij ui ) d V . x j
The second integral can be transformed into a surface integral by means of Gauss relation, which yields:
WV =
ij
ui d V x j
n u d S .
S ij j i S
Taking (8.31) into account, we obtain:

WV =
ij
ui d V x j
ij n j ui d S .
(8.36)
Equations (8.33) and (8.36) show that the total variation W of the work of volume and surface forces applied (the virtual work) is, finally:
W =
ij
ui d V , x j
t u d S + f u d V .
V
(8.37)
with
W = WS + WV =
(8.38)
Relation (8.37) expresses into the most general form the virtual work theorem for a deformable solid. In the case of small deformation assumption (6.11), and by taking into account the symmetries of ij and ij, we have:
ij
ui + ji u j = ij x j xi
u u j i + x j xi = 2 ij ij = ij ij + ji ji ,
and Relation (8.37) is written as: W =
ij ij d V .
(8.39)
The form of this relation can be transformed in the case where ij ij can be
8.3 Energy Formulation
119
expressed as the variation of a strain function ud(ij):
ij ij = ud .
(8.40)
The function ud is the strain energy per unit volume. It can also be considered as the stress function of the stress field, according to the relation:
ij =
since:
ud , ij
(8.41)
ij ij =
The volume integral is then written as:
ud ij = ud . ij
setting:
ij ij d V =
ud d V =
ud d V = U d ,
(8.42)
Ud =
ud d V ,
(8.43)
where Ud is the total strain energy. In this case, the theorem (8.39) is written as:
W = U d .
(8.44)
The virtual work is equal to the variation of the total strain energy. Usually, this latter form of the virtual work theorem is written in the equivalent form:
U = 0 ,
(8.45) (8.46)
introducing the total potential energy:

U = Ud W .
In this form, the virtual work theorem is equivalent to making the total potential energy functional stationary (i.e. extremal). For a material having a linear elastic behaviour, the strain energy is defined, according to (8.41) and (7.1), by:
i =
ud = Cij j i
i, j = 1, 2, 3, 4, 5, 6,
(8.47)
using the single subscript matrix notation for the stresses and strains. From this, we deduce that ud is a symmetric and linear quadratic form of the strains (since Cij = Cji), which, to within a constant, may be written as:
1 1 ud ( i ) = Cij j i = i i 2 2
i, j = 1, 2, . . . , 6,
(8.48)
120
or, in the two subscript tensor notation, the volume energy is: 1 ud ( ij ) = ij ij 2 The total strain energy is then written as:
Ud = i, j = 1, 2, 3 .
(8.49)
1 2
ij ij d V
i, j = 1, 2, 3 .
(8.50)
8.3.3 Dynamics of Solids

In the preceding subsection, we have considered the case of a problem of statics. In the case of the motion of a deformable solid, Expression (8.34) for the work of body forces is modified, according to Relation (8.12). Considering the results obtained previously, we easily find that the work of the body and surface forces is expressed as:
W = WV + WS =
ij ij d V +
2ui t 2
ui d V .
(8.51)
To introduce a functional, the preceding relation is integrated between two instants t1 and t2, such that the variations ui satisfy the boundary conditions:
ui = 0 on ( Su ) for t = t1 and t2 .
(8.52)
Integrating by parts shows that:

t1
t2
2ui t
2
ui d V d t =
t2 t1
Ec d t ,
(8.53)
where Ec is the kinetic energy of the solid in motion: Ec = 1 2
ui ui dV t t
i = 1, 2, 3 .
(8.54)
The integration of relation (8.51) between the instants t1 and t2 finally leads to:
t2 t1
( Ec + W ) d t =

t1
t2
ij ij d V d t .
(8.55)
This equation constitutes the variational expression, in the most general form, of the motion of a deformable solid. The integral
ij ij d V represents the energy induced during the defor-
mation of the solid or structure. Its expression depends on the behaviour law of
8.4 Variational Methods
121
the material which constitutes the solid or structure. This law is eventually a function of time. In the case in which there exists a strain energy function (8.40), Relation (8.55) has the form:
t2 t1
( Ec + W U d ) d t = 0 .

(8.56)
Usually, this relation is written as: on introducing the Lagrangian:

L = Ec + W U d = Ec U .
t2 t1
Ldt = 0 ,
(8.57)
(8.58)
Relation (8.57) constitutes the variational formulation of the problem of the motion of a deformable solid.
8.4 VARIATIONAL METHODS 8.4.1 Principle

Approximation methods are used when exact solutions of a problem of the mechanics of deformable solids cannot be derived. The general principle for approximating a solution u(M) consists to seeking the best approximation, which ( M ) , satisfying at best the different relations that the soluwe shall denote by u tion u(M) has to satisfy in the solid or structure. Among the approximation methods, the Ritz method, based on a variational approach, provides an efficient method for obtaining approximate solutions of the problem of the mechanics of deformable solids. In the displacement approach, the exact solution of the variational formulation is one which belongs to the space of admissible functions and satisfies the boundary conditions, and in the case of a problem of statics, makes the total potential energy extremal (Relation (8.45)). This relation implies that the exact solution is characterized by an absolute extremum. In the Ritz method, the approximate solution is sought as belonging to the space generated by n basis functions i(M). So, the approximate solution is written in the form:
(M ) = u
aii (M ) .
i =1
(8.59)
The functions i(M) are known functions, which are chosen a priori and the set of which constitutes a functional basis. The coefficients ai, that are to be determined,
122
are called the generalized coordinates. The basis functions i(M) have to satisfy the continuity conditions, the displacement prescribed at boundary, and to constitute a complete functional basis in order to assure convergence conditions for the approximation. The basis functions must also be derivable up to an order equal to at least that corresponding to the differential equations. The approximation sought is then the one which makes the approximate total potential energy extremal: (u (ai ) , ) = U U (8.60) according to the equation: (ai ) = 0 . U (8.61) The approximation is then characterized by a relative extremum. The approximated total potential energy thus appears in the form of a function of the coefficients ai. And the approximation sought is characterized by the n stationarity in terms of the coefficients ai, deconditions on the approximated function U duced from Relation (8.61). Thus: U = 0, i = 1, 2, . . . , n . (8.62) ai The total potential energy is a quadratic form of the coefficients ai, and it results that the stationarity conditions (8.62) lead to a system of n linear equations, which allow the n coefficients ai to be determined. In the case of the motion of a deformable solid or structure, the variational form (8.45) must be replaced by the form (8.57). In the case of the study of free vibrations, the approximated displacement field can be written by separation of the variables in the form:
(M , t ) = u ( M )ei t = u
The kinetic energy (8.54) is then written as: Ec = 1 2
aiieit .
i =1
(8.63)
i u i d V , 2u
i = 1, 2, 3 .
(8.64)
The problem of free vibrations can then be treated as a problem of statics, by simply considering the kinetic energy as an additional energy. In this case, Relation (8.57) is written as:
(U Ec ) = 0.
(8.65)
This relation leads to the usual equation of the eigenvalue problem. The natural frequencies of the deformable solid or structure are obtained by setting to zero the determinant of the coefficients ai in the equation: c (ai ) U (ai ) E = 0. ai (8.66)
Exercises
123
8.4.2 Convergence
To ensure convergence of approximated solutions as n increases, the functional basis i(M) must be complete. A functional basis is said to be complete if any function u(M) can be represented in the domain of definition, by increasing the number of generalized coordinates. That is: lim u ( M ) n =0. aii (M )
i =1 n
(8.67)
Polynomial, trigonometric, and Chebyshev functions are examples of complete functional bases. In practice, because of truncation to n terms, the basis will be said to be relatively complete. Truncation will be carried out by excluding terms of high order and will be able to represent constant or zero states of deformation. The rapidity of convergence will depend upon the way in which the chosen basis functions are appropriate for approximating the exact solution. The convergence will be the faster the better the boundary conditions are satisfied by the basis functions. The key to the Ritz method lies in the choice of functional basis. This choice is not always easy. Indeed, the stress conditions imposed on the boundary are potentially included in the expression of the total potential energy. It results from this that, in the Ritz analysis, the basis functions can satisfy only the conditions of displacements imposed on the boundaries. The reason for which it is not necessary to satisfy the conditions imposed on strains results from the fact that the stationarity conditions (8.62) minimize the divergence between the solution deduced from approximation (8.63) and the exact solution.
EXERCISES
8.1 Derive Equations (8.25) for the strain expressions in the case of cylindrical
coordinates.
8.2 Derive the fundamental equations (8.26) in cylindrical coordinates in the

case of a problem of statics. Next consider the case of a problem of dynamics.
Part III
Mechanical Behaviour of Composites Materials
This part develops the analysis of the mechanical behaviour of composite materials at the scale of the constituents (matrix and fibres). Chapter 9 considers the elastic behaviour of a unidirectional composite: the law of elasticity, the estimation of the composite moduli and the comparison of the results deduced from models and the experimental results. Composite materials with a cloth reinforcement are orthotropic materials the elastic behaviour of which is studied in Chapter 10. Laminated composite materials are constituted of successive layers the directions of which are changed from one layer to another. Thus, Chapter 11 analyzes the elastic behaviour of a unidirectional or orthotropic material off its material directions. A special attention is focused on the state of plane stress which will be applied in the laminate theory (Part IV). Lastly, Chapter 12 introduces the fundamental fracture processes induced in composite materials, and develops different fracture criteria used for evaluating the strength of laminates.
CHAPTER 9
Elastic Behaviour of a Unidirectional Composite Material
9.1 EFFECTIVE MODULI 9.1.1 The Concept of Homogenization

At a sufficiently fine scale, all the materials are heterogeneous, even the materials said to be homogeneous when we consider the level of atoms and molecules. If the usual engineering materials had to be characterized at this level of observation, the description of behaviour could not be achieved. To overcome this difficulty, engineers introduce the hypothesis of the continuity of material. This hypothesis implies the notion of statistical averaging, for which the actual constitution of the material is idealized by considering the material to be continuous. Once the model of continuity is admitted, the concept of homogeneity is deduced from it. A homogeneous material is then characterized by properties which are identical at every point of the material. At the engineering level, the character of heterogeneity appears whenever the physical or mechanical properties of a material are discontinuous functions of the point. In the case of composite materials, the properties of the materials vary in a discontinuous way at the interfaces between the various phases of materials. Each phase is assumed to be homogeneous and isotropic. In the case of a phase 1 dispersed in a phase 2 (Figure 9.1), there exists in general a characteristic dimension of the heterogeneity. For example, in a fibre composite material, this dimension is the average distance between fibres. Furthermore, there generally exists a scale of size at which the properties of the material can be averaged to a good approximation. This means that, in this case, the properties measured on a sample of size are independent of the place ( of the point ) within the material at which the sample was taken. In terms of such a concept, the material an then be considered as being effectively homogeneous,
128
Chapter 9 Elastic Behaviour of a Unidirectional Composite
phase 1 phase 2
characteristic dimension
averaging dimension
FIGURE 9.1. Homogenization of a heterogeneous material.
and the problems of structure designing can be solved by considering the average properties measured at the scale . In the case where there exists such a scale (intermediate between the scale of the constituents and the scale of the structure), it is said that the material can be homogenized. One then speaks of macroscopic homogeneity or material homogeneity (as opposed to constituent scale). The concept of considering as homogeneous a heterogeneous material is called the homogenization concept. The alternative to this concept would be to take account of each region of homogeneity, and to analyze the continuity of stresses and displacements through each interface. If such an approach is conceivable in principle, it is still inaccessible today in practice, on account of the high number of interfaces that must be considered (several thousand to several million). Nevertheless, the decrease in the times of calculations and the increase in the size of computer memories are making possible an approach at an increasingly finer scale.
9.1.2 Homogenized Moduli

The concept of homogenization having been introduced, it is now possible to express the homogenized mechanical properties of a heterogeneous material. These properties are determined for an element of volume V and size . This volume element is called representative volume element of the material. When stress and strain conditions are imposed on the boundary of this volume element, the average stress (the stress matrix) applied on the representative volume element is defined by:
9.2 Hookes Law for a Unidirectional Composite
129
i =
and the average strain is:
1 V
i ( xk ) d V ,
i = 1, 2, . . . , 6,
(9.1)
j =
1 V
j ( xk ) d V ,
i = 1, 2, . . . , 6,
(9.2)
where i and j are the components of the stress and strain matrices at point xk, and dV is the volume of the element surrounding the point xk. These relations are quite general and allow us to derive the effective stiffness Cij and compliance Sij constants using the expressions:
i = Cij j ,
and
i, j = 1, 2, . . . , 6 ,
i, j = 1, 2, . . . , 6 .
(9.3)
i = Sij j ,
(9.4)
It is according to this concept of homogenization that in the following the stiffness and compliance constants will be considered. Thus, in order to determine the homogenized properties of a heterogeneous material, it is necessary to calculate the average stress and strain over the representative element by means of Equations (9.1) and (9.2), and then to deduce the stiffness and compliance constants according to Equations (9.3) and (9.4). If this problem appears to be simple in principle to solve, it is particularly complex in practice. In fact, in order to apply Expressions (9.1) and (9.2), it is necessary first to find the exact solutions of the stress and strain fields, i(xk) and j(xk), at each point of the heterogeneous material. These exact solutions can be obtained only in the case of simple and idealized geometric models, and somewhat distant from reality.
9.2 HOOKES LAW FOR A UNIDIRECTIONAL COMPOSITE 9.2.1 Constitution of a Unidirectional Composite Material
A unidirectional composite is constituted of parallel fibres arranged in a matrix (Figure 9.2a). This type of material forms the basic configuration of fibre composite materials, hence the importance of studying it. The elementary cell of such a material can be considered, to a first approximation, as constituted of fibre embedded in a cylinder of the matrix with a circular base (Figure 9.2b), or, better,
130
1, L 1, L 2, T
3, T
(a)
(b)
FIGURE 9.2. Unidirectional composite.
a hexagonal base. We shall return to this aspect in Section 9.3.1. This cell has an axis of revolution, which we shall note axis 1. This direction parallel to the fibres is called the longitudinal direction, and for this reason the axis 1 is also denoted as the L-axis. Every direction normal to the fibres is called a transverse direction, and the composite is considered as being transversely isotropic: it is isotropic in the plane normal to the direction 1. The transverse plane will be described by the directions 2 and 3, also denoted by T and T , these directions being equivalent.
9.2.2 Stiffness and Compliance Matrices

The elastic behaviour of a unidirectional composite material can be described by introducing (Chapter 7) either the stiffness constants Cij, or the compliance constants Sij. Taking into account the results established in Chapter 7, Relation (7.17), the Hookes law is written in one of the two matrix forms:
1 C11 C12 C12 C C22 C23 2 12 3 C12 C23 C22 = 0 0 0 4 5 0 0 0 6 0 0 0 or the inverse form: 0 0 0 1 C C 23 ) 2 ( 22 0 0 0 0 0 0 C66 0 0 1 0 2 0 3 , 0 4 0 5 6 C66
(9.5)
9.3 Engineering Constants
131
1 S11 S 2 12 3 S12 = 4 0 5 0 6 0
S12 S22 S23 0 0 0
S12 S23 S22 0 0 0
0 0 0 0 0 0 2 ( S22 S23 ) 0 S66 0 0 0
0 1 0 2 0 3 . 0 4 0 5 S66 6
(9.6)
The stiffness and compliance matrices are inverses of each other, and the elastic behaviour of a unidirectional composite material is thus characterized by 5 independent coefficients: C11, or S11, S12, S22, S23, S66. C12, C22, C23, C66,
9.3 ENGINEERING CONSTANTS

The engineering constants are the Youngs moduli, the Poisson ratio and the shear moduli. These constants are measured in simple tests such as uniaxial tension or shear tests. These constants thus correspond to a more practical use than the stiffness or compliance constants. In general, these tests are implemented by imposing a known stress field, and then measuring the strain field. It follows from this that the compliance constants are related to the engineering constants by relations simpler than those expressing the stiffness constants. Hereafter, we establish these various relations by considering different fundamental tests.
9.3.1 Longitudinal Tensile Test

In a longitudinal tensile test, all the stresses are zero except for the stress 1:
1 0, i = 0,
i = 2, 3, . . . , 6.
Considering the stiffness constants, the elasticity equations (9.5) may be written as:
1 = C111 + C12 2 + C12 3 ,

0 = C121 + C22 2 + C23 3 , 0 = C121 + C23 2 + C22 3 ,
4 = 5 = 6 = 0.
132
From these relations, we deduce:
2 = 3 =
and
C12 1, C22 + C23 (9.7)

2 C12 1. C22 + C23
1 = C 2 11
Thus, we obtain the longitudinal Youngs modulus EL and the Poisson ratio LT for longitudinal tension: EL = C11
2 2C12 , C22 + C23
LT =
C12 . C22 + C23
(9.8)
As functions of the compliance constants, the elasticity equations (9.6) for a longitudinal tensile test may be written as:
1 = S11 1, 2 = 3 = S12 1 , 4 = 5 = 6 = 0.
Whence: EL = 1 , S11 S12 . S11
LT =
(9.9)
9.3.2 Transverse Tensile Test

In transverse tension, for example in the 2-direction, the imposed stress field is:
2 0, i = 0,
i 2.
The elasticity equations, considering the stiffness constants, in this case are written as: 0 = C111 + C12 2 + C12 3 ,
2 = C121 + C22 2 + C23 3 ,

0 = C121 + C23 2 + C22 3 ,
4 = 5 = 6 = 0.
From which we deduce:
133
1 = 3 =
C12 ( C23 C22 )

2 C11C22 C12 2 C11C23 C12 2 C11C22 C12
2 ,
(9.10)
2 ,
2 2 C12 C22 2C23 ) + C11C23 ( 2 = C22 + 2. 2 C C C 12 11 22
Thus, we derive the expressions for the transverse Youngs modulus ET and the Poisson ratios 21 and 23, denoted respectively by TL and TT :
EL = C22 +
2 2 C12 ( C22 2C23 ) + C11C23 2 C12 C11C22
, (9.11)
TL = TT =
C12 ( C23 C22 )

2 C12 C11C22 2 C12 C11C23 2 C12 C11C22
Introducing the compliance constants, the elasticity equations for transverse tension are written as: 1 = S12 2 ,
2 = S22 2 , 3 = S23 2 , 4 = 5 = 6 = 0.
Thus:
1 =
and
S12 2 , S22 1 , S22
3 =
S23 2, S22 S12 , S22
2 =
1 2, S22 S23 . S22
(9.12)
ET =
TL =
TT =
(9.13)
These relations, when compared with Expressions (9.9), show that the coefficients EL, ET, TL and LT are connected by the relation:
LT
EL
TL
ET
(9.14)
9.3.3 Longitudinal Shear Test

A longitudinal shear test corresponds to one of the stress states :
134
5 0, 6 0, or i = 0 si i 5, i = 0 In the second case, the elasticity equations are:
si i 6.
1 = 2 = 3 = 4 = 5 = 0, 6 = C66 6 .
Thus, we deduce the longitudinal shear modulus GLT as: GLT = C66 or GLT = 1 . S66
(9.15)
(9.16)
Because the directions T and T are equivalent, we have:

GLT = GLT = C66 .
(9.17)
9.3.4 Transverse Shear Test

In a transverse shear test, the stress field is expressed by:
4 0, i = 0 if i 4.
Whence the relations:
i = 0 4 =
if i 4,
1 ( C22 C23 ) 4 . 2
The transverse shear modulus GTT is thus written as:

GTT = 1 ( C22 C23 ) 2 or GTT = 1 . 2 ( S 22 S 23 )
(9.18)
The transverse shear modulus GTT is related to the transverse Youngs modulus ET and the Poisson ratio TT by the expression:
GTT = ET . 2 (1 + TT )
(9.19)
9.3.5 Lateral Hydrostatic Compression

A lateral hydrostatic compression test without longitudinal deformation also allows a simple characterization of materials. In such a test, the applied stress and strain fields are such that:
135
2 = 3 = p, 4 = 5 = 6 = 0, 1 = 0, 1 0,
where p is the applied hydrostatic pressure. The elasticity equations are then written as: 1 = C12 2 + C12 3 , p = C22 2 + C23 3 , p = C23 2 + C22 3 , 4 = 5 = 6 = 0. From these equations, we deduce:
p = ( C22 + C23 ) 2 ,
2 = 3,
1 = 2C12 2 .
The surface dilatation es is: es = 2 + 3 = Whence: p= 1 ( C22 + C23 ) es . 2 (9.20) 2 p. C22 + C23
Thus, we deduce the lateral compression modulus KL without longitudinal deformation as:
KL =
1 ( C22 + C23 ) . 2
(9.21)
Similarly as function of compliance constants, we obtain: KL = 1

2 S12 2 S22 + S23 2 S11
(9.22)
The modulus KL is related to the longitudinal Youngs modulus EL and the Poisson ratio LT by the expression:
KL = C11 EL
2 4 LT
(9.23)
136
9.3.6 Moduli as Functions of the Stiffness and Compliance Constants

Here we resume a synthesis of the results obtained in the preceding subsections. In these subsections, we have introduced the engineering constants, measured in simple stress and strain states: EL and LT, the Youngs modulus and the Poisson ratio, measured in a longitudinal tensile test ; ET, TL, TT', the Youngs modulus and the Poisson ratios, measured in a transverse tensile test ; GLT and GTT', the shear moduli measured respectively in longitudinal and transverse shear tests ; KL, the hydrostatic compression modulus measured in a lateral hydrostatic compression without longitudinal deformation. We have seen (Subsection 9.2.2) that the mechanical behaviour of a unidirectional material is described by 5 independent constants. Among the engineering constants considered, only 5 moduli are then independent. For example, we have the relations: ET = 2
2 1 1 + + 2 LT EL 2 K L 2GTT
1 2 2 1 2 LT 2 ET 2 K L EL
(9.24)
GTT =
(9.25)
TT =
ET 1, 2GTT ET . EL
(9.26)
TL = LT
(9.27)
The engineering constants usually used in practice are: EL, ET, LT, GLT and GTT'. Furthermore, we shall see that the modulus KL can be easily estimated by analytical relations as function of the characteristics of the constituents of the composite, whence the interest in KL. The expressions for the moduli as functions of the stiffness constants are the following ones:
137
EL = C11 ET = C22 +
2 2C12 , C22 + C23
LT =
C12 , C22 + C23 , (9.28)

2 C12 2 C12
2 2 C12 ( C22 2C23 ) + C11C23 2 C11C22 C12
TL =
C12 ( C23 C22 )

2 C11C22 C12
TT =
C11C23 C11C22
GLT = C66 ,
GTT =
1 ( C22 C23 ) , 2
KL =
1 ( C22 + C23 ) . 2
The expressions for the moduli as functions of compliance constants are: EL = ET = GLT = KL = 1 , S11 1 , S22 1 , S66
LT = TL =
GTT = 1
S12 , S11 S12 , S22
TT =
S23 , S22 (9.29)
1 , 2 ( S22 S23 ) .
S2 2 S22 + S23 2 12 S11
9.3.7
Stiffness and Compliance Constants as Functions of Moduli
The inverse relations, allowing us to determine the stiffness and compliance constants as functions of the engineering constants, are obtained without any difficulty. For the stiffness constants, they are written as:
2 C11 = EL + 4 LT KL ,
C12 = 2 K L LT , C22 = GTT + K L , C23 = GTT + K L , C66 = GLT .
(9.30)
138
For the compliance constants, the relations are:

S11 = S22 S66 1 , EL 1 = , ET 1 = . GLT S12 = S23 =
LT
ET
, , (9.31)
TT
ET
9.3.8 Restrictions on the Engineering Constants

The elasticity constants have values which have to be in agreement with the basic physical principles. For example, tension on a test specimen can imply only a positive extension in the direction of tension. Lateral hydrostatic compression can induce only a lateral contraction. Such considerations induce the following conditions for the values of the moduli: the moduli EL, ET, GLT, GTT' and KL are positive:
EL , ET , GLT , GTT , K L > 0,
(9.32)
from Expression (9.24), it is deduced:

2 LT
ET < 1, EL
or
LT <
EL , ET
(9.33)
Expression (9.33) combined with (9.27) leads to:
TL <
ET , EL
(9.34)
Expression (9.19) of GTT' similarly leads to:

1 TT 1.
(9.35)
9.4 THEORETICAL APPROACHES FOR EVALUATING THE ELASTICITY CONSTANTS 9.4.1 Problem and Different Approaches
The problem of determining the elasticity constants of a unidirectional composite consists in searching for the expressions of these constants (5 independent constants) as functions of the mechanical and geometric properties of the constituents: the elasticity moduli of the matrix and fibres, the volume fraction of the
9.4 Theoretical Approaches for Evaluating the Elasticity Constants
139
fibres, the length of the fibres, etc. This section will be restricted to the study of composites with continuous fibres. The mechanical properties for the fibres and the matrix will be described by their elasticity moduli (Youngs moduli and Poisson ratios), denoted respectively by Ef, f, Em, m. Solving the problem is not simple, and also the solution is not unique. In this section, we present a short synthesis of the problem and its complexity. Readers interested by a more extended analysis should refer to the bibliographical and critical work on this subject carried out by C.C. Chamis and G.P. Sendeckyj [4]. The complexity of the problem can be simply illustrated by considering the problem of the description of the arrangement of fibres in a unidirectional composite. In fact, this material is made of parallel continuous fibres whose proportion is imposed and characterized by the volume fraction Vf of the fibres. In practice, there does not exist only one possibility for the arrangement of the fibres, but an infinite number. For example, if the fibres are distributed regularly, they will be able to be distributed in a hexagonal array (Figure 9.3), a square array (Figure 9.4), a staggered square array (Figure 9.5), etc. The corresponding elementary cells are reported in the same figures. The distributions of the fibres in a square array and in a staggered square array have the same elementary cell. These two distributions differ by a rotation through 45 from the principal directions of the applied stresses. In practice, during the fabrication of the unidirectional composite, the fibres are most often distributed randomly, rather in a regular arrangement. The actual distribution observed can then be illustrated by the fibre arrangement shown in Figure 9.6: some fibres are completely surrounded by matrix, although others might touch. The theoretical analysis would then have to take that into account. In practice, the actual solution will be intermediate between the solution found in the case where the fibres are isolated from each other, and the solution which would be obtained in the case in which all the fibres would be in contact. All the previous considerations thus show the difficulties which arise in the theoretical approaches to the determination of the elasticity moduli. The methods used can be classified into three types: the evaluation of bounds (bonding expressions) using energy variational theorems, the determination of exact solution in the case of simple arrangement of the fibres, expressions derived from semiempirical approaches.
FIGURE 9.3. Hexagonal array and the elementary cell.
140
FIGURE 9.4. Square array and the elementary cell.
FIGURE 9.5. Staggered array and the elementary cell.
In the evaluation of bounding expressions, energy variational theorems (such as the total potential energy theorem, Section 8.3) are used to determine upper and lower bounds of the elasticity moduli. However, these approaches generally lead to bounds that may not be sufficiently close to be used in practice. The determination of exact solutions is usually complex, and can be achieved only in simplified schemes: simplified geometry of the elementary cell, simple conditions prescribed on the boundaries of the cell, etc.). The solution can be derived either by an analytical approach or using a numerical method such as the finite element analysis. Last, semi-empirical methods allow us to obtain relations of practical use, but do not allow an actual prediction of the properties. They therefore remain descriptive. In the following subsections, we develop some elements of these various approaches.
FIGURE 9.6. Schematic representation of the actual arrangement of fibres.
141
9.4.2 Bounds on the Elasticity Moduli

The determination of bounds on the elasticity moduli uses the energy variational theorems. In a displacement approach, the total potential energy theorem (8.46) allows us to find upper bounds, when the complementary potential energy theorem, in a stress approach, allows us to obtain lower bounds [5]. Among the works of the literature developed in this way, Hashin [6] and Hill [7] studied the case of fibres of different diameters distributed randomly in the matrix (Figure 9.7), but with a given volume fraction. Hashin and Rosen [8] developed analogous works in the case of fibres of identical diameters distributed in a hexagonal arrangement (Figure 9.8). The expressions obtained are expressed as functions of the lateral compression moduli (without longitudinal deformation) Km and Kf, of the matrix and fibres, respectively. These moduli are related to the bulk moduli (7.41) (km, kf) and the shear moduli (Gm, Gf) by the expressions: Ki = ki + Gi , 3 i = m, f. (9.36)
The moduli (km, kf) and (Gm, Gf) are themselves expressed (Table 7.1) as functions of the Youngs moduli (Em, Ef) and Poisson ratios (m, f) of the matrix and fibres by: Ei (9.37) ki = , i = m, f, 3 (1 2 i )
Gi = Ei , 2 (1 + i ) Ei i = m, f.
(9.38)
So, finally:
Ki = 2 (1 2 i )(1 + i ) , i = m, f.
(9.39)
FIGURE 9.7. Random arrangement of fibres.
142
FIGURE 9.8. Hexagonal arrangement of fibres of the same diameter.
Considering the notations introduced in Section 9.3, the bounds for the five independent moduli, derived by Hashin [6] and Hill [7] are given by the expressions: Km + Vf K L Kf + 1 Vf 1 Vf + K m K f K f + Gf , (9.40)
1 1 Vf + K f K m K m + Gm
Gm +
Vf 1 K m + 2Gm + (1 Vf ) 2Gm ( K m + Gm ) Gf Gm GTT Gf + 1 K f + 2Gf + Vf 2Gf ( K f + Gf ) Gm Gf 1 Vf , 1 Vf + Gm Gf 2Gf 1 Vf ,
(9.41)
Gm +
1 1 Vf + Gf Gm 2Gm
Vf
GLT Gf +
(9.42)
Vf (1 Vf ) EL Ef Vf Em (1 Vf ) Vf (1 Vf ) , 2 Vf 1 Vf 1 1 Vf Vf 1 4 ( ) f m + + + + Km Kf Gm Kf K m Gf Vf (1 Vf ) LT f Vf m (1 Vf ) Vf (1 Vf ) . Vf 1 Vf 1 1 Vf Vf 1 1 1 + + + + ( f m ) Km Kf Gm K K Gf K K f m f m
(9.43)
(9.44)
143
9.4.3 Exact Solutions

The determination of exact solutions is implemented in the case of particular arrangements of fibres, by one of the classical approaches of the mechanics of deformable solids (Chapter 8). In this section, we consider the analytical approach of displacement type applied to the analysis of the problem of a cylindrical elementary cell submitted to a tension state. The geometric model used for the elementary cell (Figure 9.9) is a cylindrical fibre of radius rf, embedded in a matrix cylinder of radius rm. The fibre and matrix radii are related to the volume fraction Vf of the fibres by the relation:
Vf = rf2
2 rm
(9.45)
9.4.3.2 Preliminary Problem: Material Cylinder Submitted to a Uniform Tension

We have first to solve the problem of a cylinder of an isotropic homogeneous material submitted to a uniform tension (Figure 9.10) in the direction of the cylinder axis. The radial symmetry of the problem requires the use of cylindrical coordinates (Subsection 8.2.3). We consider the following displacement field in the cylinder: B ur = Ar + , r u = 0, (9.46) w = Cz ,
defined at every point M of the cylinder, except in the vicinity of the axis (r = 0). It is easy to show that this field is solution of the elasticity problem under consideration if r 0. In fact, according to (8.25), the strain field at point M is written as:
fibre
rf
rm
FIGURE 9.9. Cylindrical elementary cell of a unidirectional composite.
144
1 M 2
FIGURE 9.10. Cylinder submitted to a uniform tension.
rr ( M ) = 0 0
0 0 , zz
(9.47)
with: ur B = A 2 , r r 1 B = ur = A + 2 , r r w zz = = C. z
rr =
(9.48)
We find without difficulty, starting from these relations, that the compatibility conditions (6.18) are satisfied. The stress field is next determined considering the elasticity relations (7.24). Whence:
rr (M ) = 0 0 0
0 0 , zz
(9.49)
with:
rr = 2 K A + C (1 2 ) = 2 K A + C + (1 2 ) zz = 2 K 2 A + (1 ) C ,

B , r2 B , r2
(9.50)
By introducing the lateral compression modulus defined in (9.39). These expressions verify the equilibrium equations (8.26). From this, it results that the displaycement field introduced in (9.46) is the solution of the elasticity problem considered. Constants A, B and C are to be determined considering the boundary conditions imposed.
145
9.4.3.3 Fibre Embedded in a Matrix Cylinder Subjected to a Traction

We now study the problem of the cylindrical cell (Figure 9.9) submitted to a uniaxial tension in the fibre direction. Two types of conditions can be prescribed: either tensile strains are prescribed: or tensile stresses are prescribed:
zz = , zz = .
(9.51) (9.52)
We study the case in which tensile strains are prescribed. The case of prescribed tensile stresses would lead to similar results. As for displacement fields in the matrix and fibre, we introduce fields of the type (9.46). However, the displacement in the fibre being finite when r tends zero requires that the coefficient B be zero in the fibre. Thus, we consider as displacement fields: in the fibre: urf = Af r , u f = 0, (9.53) w f = Cf z , in the matrix:
urm = Am r + u m = 0, w m = C m z. The strains in the matrix and fibre are respectively: Bm , r (9.54)
zzm = Cm
and
zzf = Cf .
The condition (9.51) of prescribed longitudinal strains ( zzf = zzm = ) leads to:
Cf = Cm = .
(9.55)
The other constants Af, Am and Bm are determined considering: the conditions prescribed at the boundary ( r = rm ):
rrm = 0,
the continuity conditions at the fibre-matrix interface ( r = rf ): for the displacements: urf = urm , u zf = u zm (continuity satisfied by (9.55)), for the radial stresses:
(9.56)
(9.57)
rrf = rrm .
(9.58)
146
Having derived the constants Af, Am and Bm, the elasticity moduli will be deduced according to Relations (9.3). In the case of the tension under consideration, we shall therefore obtain: the longitudinal modulus EL: EL = E11 = thus:
EL =
1
2 rm
zz ,
zz ( r ) d S
(9.59)
where S is the area of the cross section of the cell; the Poisson ratio LT : urm (r = rm ) . rm
LT = 12 =
(9.60)
The conditions prescribed at the cell boundary (9.56), at the interface (9.58) and the preceding expressions of moduli show the necessity of first determining the expressions for the stresses rr and zz. These expressions are deduced from Relations (9.50). We obtain: in the fibre: (9.61) rrf = 2 K f ( Af + f ) ,
zzf = 2 K f 2 f Af + (1 f ) .
in the matrix:
(9.62)
rrm = 2 K m Am + m (1 2 m )
zzm
Bm , r2 = 2Km 2 m Am + (1 m ) .
(9.63)
(9.64)
The conditions (9.57) and (9.58) of continuity at the fibre-matrix interface (r = rf) require that: B Af rf = Am rf + m , (9.65) rf B K f ( Af + f ) = K m Am + m (1 2 m ) m , rf2 and the condition (9.56) at the cell boundary ( r = rm ) leads to: B Am + m (1 2 m ) m =0. 2 rm (9.66)
(9.67)
The three preceding conditions allow us to derive the constants Am, Bm and Af. We obtain:
147
Am = A ,
2 2 1 rm m rm Af = A 1 + + , 1 2 m r 2 1 2 m r 2 f f
(9.68)
Bm = with:
A + m 2 rm , 1 2 m
m
A= Kf
(1 Vf ) +
Gm K m Vf 1 1 Vf + + K m Gm Kf
Vf . (9.69)
The constants being obtained, the longitudinal Youngs modulus is deduced from Equation (9.59). So in the case of cylindrical coordinates:
EL = 1
2 rm

=0
rm r =0
zz (r ) r dr d ,
rm rf
or
EL =
2 2 rm
rf 0
r zzf (r ) dr +
r zzm (r ) dr .
(9.70)
According to (9.62), (9.64), (9.68) and (9.69), the stresses zz are written as: in the fibre: with 1 1 C zf = 1 f + 2 f A + ( A + m ) , 1 2 m Vf in the matrix:
zzf (r ) = 2 K f C zf ,
(9.71)
zzm (r ) = 2 K mC zm ,
with
C zm = 1 m + 2 m A .
(9.72)
Substituting Relations (9.71) and (9.72) into Expression (9.70) of the modulus, we obtain: EL = 2 K f C zf Vf + 2 K mC zm (1 Vf ) . The development of this relation finally leads to the expression of the longitudinal modulus as: 4 ( f m ) Vf (1 Vf ) EL = Ef Vf + Em (1 Vf ) + . Vf 1 Vf 1 + + Km Kf Gm
2
(9.73)
148
Similarly, considering Relation (9.60) leads to the expression of the Poisson ratio:
LT = f Vf + m (1 Vf ) +
( f m )
1 1 Vf (1 Vf ) K m Kf . Vf 1 Vf 1 + + Km Kf Gm
(9.74)
Relations so obtained (9.73) and (9.74) are identical to the lower bounds of the longitudinal modulus (9.43) and Poisson ratio (9.44). Moreover, numerical applications, carried out using Expressions (9.73) and (9.74), show that these expressions are practically identical to the mixture laws that will be established simply in Subsections 9.4.4.1 and 9.4.4.3.
9.4.3.4 Other Moduli

In the same way, it is possible to find exact solutions of two other elasticity problems of the cylindrical cell: longitudinal shear of the cell (Subsection 9.3.3), lateral hydrostatic compression of the cell (Subsection 9.3.5). Solving the first problem leads to the expression for the longitudinal shear modulus: GLT = G12 = Gm Gf (1 + Vf ) + Gm (1 Vf ) , Gf (1 Vf ) + Gm (1 + Vf ) (9.75)
whereas the second problem leads to the expression of the lateral compression modulus: K L = Km + 1 Vf + 1 4G kf km + 3 ( Gf Gm ) km + 3 m 1 Vf . (9.76)
Relations (9.73) to (9.76) were established by Hill [7] and Hashin [9]. Relation (9.75) obtained for the longitudinal shear modulus is identical to the lower bound (9.42). Similarly, Expression (9.76) is identical to the lower bound of KL (9.40). Having obtained four independent moduli, it remains to derive a fifth one, for example the transverse shear modulus GTT ' = G23, determined in a transverse shear test (Subsection 9.3.4). However, no exact analytical solution has been found for the transverse shear problem of the elementary cylindrical cell. No expression of the modulus GTT ' , analogous to the preceding expressions found for the other moduli, can thus be proposed in this context. This statement leads to considering another approach described in the following subsection.
149
1, L
fibre matrix equivalent homogeneous material FIGURE 9.11. Cylindrical model with three phases.
9.4.3.5 Model for Estimating the Transverse Shear Modulus

The method used to overcome the difficulties reported in the preceding subsection consists in considering a cylindrical model with three phases (Figure 9.11). In this model, the elementary cylindrical cell is surrounded by a cylinder of a larger size, constituted by the equivalent homogenized material having the effective properties of the composite material. The external boundary of this cylinder is submitted to the conditions of a transverse shear test. Such a model was developed by Christensen and Lo [5, 10], starting from the works carried out by Hermans [11]. This model leads to the following expression of the transverse shear modulus:
Vf GTT = Gm 1 + . 7 k G + G m m m 3 1 Vf ) + Gf Gm 2km + 8 Gm ( 3
(9.77)
This expression is identical to the lower bound (9.41) of GTT .
9.4.4 Simplified Approaches

Simplified and practical expressions of the moduli can be obtained by considering very simplified approaches to the mechanical behaviour of the elementary cell of a unidirectionl composite material. We develop these expressions in the present subsection.
9.4.4.1 Longitudinal Youngs Modulus

The longitudinal Youngs modulus is determined in a longitudinal tensile test (Figure 9.12). The simplifying hypothesis is to assume a uniform and identical
150
2, T
lt 2
1
lt
matrix fibre matrix

l
1, L
FIGURE 9.12. Simplified representation of longitudinal tension.
elongation of the fibre and the matrix. If l is the extension of the cell (identical to that of the fibre and matrix), the longitudinal strain imposed on the cell is:
1 =
l , l
where l is the length of the cell under consideration. The identity of the longitudinal strains in the fibre and in the matrix imposes: f = m = 1 . (9.78) If the fibre and the matrix have a linear elastic behaviour, the stresses in the fibre and in the matrix are expressed by:
f = Ef 1, m = Em1.
The resultant load applied to the composite is: F1 = f Sf + m Sm ,
(9.79)
where Sf and Sm are respectively the areas of the cross sections of the fibre and matrix. If S is the area of the cross section of the average cell, the average stress 1 = F1/S is written as:
1 = f Vf + m (1 Vf ) .
(9.80)
This average stress is related to the strain of the cell by the longitudinal Youngs modulus: 1 = EL1 . (9.81) Combining Relations (9.79) to (9.81) leads to the expression of the longitudinal Youngs modulus: (9.82) EL = Ef Vf + Em (1 Vf ) . This expression is known as the law of mixtures for the longitudinal Youngs modulus in the direction of the fibres. This law of mixtures is illustrated in
151
Ef
Longitudinal modulus EL
Em Fibre volume fraction Vf

FIGURE 9.13. Law of mixtures for the longitudinal Youngs modulus.
Figure 9.13. The variation of the Youngs modulus is linear between the value Em of the modulus of the matrix and the value Ef of the modulus of the fibres, when the volume fraction Vf of the fibres varies from 0 to 1.
9.4.4.2 Transverse Youngs modulus

The transverse Youngs modulus is determined in a transverse tensile test where the composite material is loaded in the direction normal to the fibres. A simplified expression for this modulus can be obtained in a simple-minded twodimensional scheme. In such a scheme, a slab of composite material, of given thickness, is considered as being constituted (Figure 9.14) of successive layers having alternatively the properties of the fibres and the matrix. The height of the layers must simply satisfy: Vf = hf hf + hm and 1 Vf = hm . hf + hm (9.83)
The load F2 applied in the transverse direction is transmitted in both the fibres and the matrix, imposing equal stresses in the fibres and the matrix:
m = f = 2 .
(9.84)
Thus, the respective strains in the fibres and the matrix in the transverse direction are given by:
f =
2
Ef
m =
2
Em
(9.85)
152
2, T
2, T
2
hm 2
hf hm 2
matrix matrix fibre
matrix fibre
FIGURE 9.14. Layers model of a unidirectional composite.
The transverse extension of the elementary cell results from the cumulative elongations in the fibre and the matrix. Thus:
l2 = f hf + m hm ,
and the transverse strain is written as:
2 =
Whence:
l2 hf hm = f + m . hf + hm hf + hm hf + hm
2 = f Vf + m (1 Vf ) .
(9.86)
This strain is related to the transverse stress applied to the cell by the transverse modulus: 2 = ET 2 . (9.87) Combining Expressions (9.85) to (9.87) leads to the expression of the transverse Youngs modulus: 1 V 1 Vf = f + . (9.88) ET Ef Em This expression is known as the inverse law of mixtures and can be rewritten as a dimensionless relation, referring the transverse modulus of composite to the Youngs modulus of the matrix, in the form: ET = Em 1 E 1 + m 1 Vf Ef . (9.89)
Table 9.1 gives the values of ET /Em for three values of the ratio Ef /Em. It is to be noted, for example, that in the case of the ratio Ef /Em = 100, a fibre volume of 50 % is needed in order to obtain a transverse Youngs modulus twice as much as
153
TABLE 9.1. Values of ET /Em for various values of the ratio Ef /Em and of the fraction Vf of fibres. Vf Ef /Em 5 10 100 0 1 1 1 0,2 1,19 1,22 1,25 0,4 1,47 1,56 1,66 0,5 1,67 1,82 1,98 0,6 1,92 2,17 2,46 0,7 2,27 2,70 3,25 0,8 2,78 3,57 4,80 0,9 3,57 5,26 9,17 0,95 4,17 6,90 16,81 1 5 10 100
the modulus of the matrix. This result shows that the fibres weakly contribute to the transverse Youngs modulus. The variation of the modulus ET as function of the volume fraction of the fibres is reported in Figure 9.15 for three values (5, 10, 100) of the ratio Ef /Em.
9.4.4.3 Longitudinal Poissons ratio

To determine the longitudinal Poissons ratio LT, we consider again the preceding scheme in which the unidirectional composite is described by successsive layers. The coefficient LT is determined in a longitudinal tensile test, illustrated in Figure 9.16. This scheme differs from that of Subsection 9.4.4.1 in the model of the elementary cell: cylindrical model in Figure 9.12 and layers model in Figure 9.16.
20
Transverse Youngs modulus EL / Em
16
Ef = 100 Em
12
Ef = 10 Em Ef =5 Em
0.2
0.4
0.6
0.8
Fibre volume fraction Vf

FIGURE 9.15. Variation of the modulus ET as function of the volume fraction Vf of fibres.
154
2, T
lt 2 hm 2 hf hm 2 l
matrix fibre matrix
1, L
FIGURE 9.16. Layers model for a longitudinal tension.
As in Subsection 9.4.4.1, fibre and matrix are submitted to identical longitudinal deformations (9.78). Thus, the transverse strains in the matrix and in the fibres are written as:
2m = m1, 2f = f 1.
The transverse extension of the elementary cell is:
lt = m1hm f 1hf ,
and the transverse strain is given by:
2 =
lt = m (1 Vf ) + f Vf 1 . hf + hm
Whence the expression for the Poisson ratio:
LT = f Vf + m (1 Vf ) .
(9.90)
This expression is the law of mixtures for the longitudinal Poisson ratio. The variation of LT as function of volume fraction of fibres is linear from m to f. In practice, the values of m and f are quite close (about 0.3). The Poisson ratio LT will thus also stay close to this value.
9.4.4.4 Longitudinal Shear Modulus

The longitudinal shear modulus GLT is determined in a longitudinal shear test illustrated in Figure 9.17, using again the layers model of the unidirectional composite. The longitudinal shear stresses in the fibre and in the matrix are equal, as a result of the longitudinal shear stresses applied on the cell. The shear strains
155
2, T
hm 2 hf hm 2
matrix
fibre matrix
1, L
FIGURE 9.17. Layers model of a longitudinal shear test.
of the fibre and the matrix are thus expressed as follows:
f =
Gf
and
m =
Gm
(9.91)
The shear deformations induced in the fibre and the matrix (Figure 9.18) are:
f = hf f
and
m = hm m .
The total deformation of the cell (Figure 9.18) is:
= f + m = hf f + hm m .
And the shear angle of the cell is determined by the expression:
hf + hm
= f Vf + m (1 Vf ) .
(9.92)
This shear angle is related to the shear stress by the longitudinal shear modulus
matrix fibre matrix
f m 2
FIGURE 9.18. Shear deformations of the matrix and fibre.
156
GLT according to the relation:
GLT
(9.93)
Combining Expressions (9.91) to (9.93), we obtain: 1 V 1 Vf = f + . GLT Gf Gm (9.94)
This expression has a form identical to that (9.88) obtained for the transverse Youngs modulus. The same considerations can be transposed here.
9.4.5 Halpin-Tsai Equations

Considering the results obtained in the theoretical analyses (Subsections 9.4.3 and 9.4.4), Halpin and Tsai [12] proposed equations that are both general and simple in formulation. The moduli of a unidirectional composite are expressed by: the law of mixtures for the modulus EL and the coefficient LT :
EL = Ef Vf + Em (1 Vf ) ,
LT = f Vf + m (1 Vf ) ,
and the following general expression for the other moduli: M 1 + Vf = , 1 Vf Mm an expression in which: the coefficient is given by: (9.95)
(Mf (Mf
M m ) 1 , Mm ) +
(9.96)
M is the modulus under consideration: ET, GLT ou TT ; Mf is the corresponding modulus of fibres: Ef, Gf,or f ; Mm is the modulus of the matrix: Em, Gm, or m. The factor is a measure of the fibre reinforcement and depends on the geometry of the fibres, the arrangement of the fibres and the type of test (thus on the modulus considered). If the simplicity of the relations is somewhat seductive, the application to designing is actually only theoretical. In fact, the difficulty of using Relation (9.95) stays in the determination of the appropriate values of . The values of this
9.5 Numerical Values of the Moduli
157
factor can be determined only by comparison to an analytical or numerical solution (we are then reduced to solve the elasticity problem), or by adjustment with experimental results. For example, Halpin and W. Tsai obtained an excellent agreement with the results obtained by Adams and Doner [13, 14] by a finite difference method applied to the case of cylindrical fibres distributed in a square arrangement and for a fibre volume fraction of 0.55, by taking: = 2 for determining the modulus ET, = 1 for determining the modulus GLT. A more extended discussion related to the comparison between Expression (9.95) and various exact (analytical or numerical) solutions is developed in Reference [12]. Finally, the results obtained show that if Relation (9.95) is well adapted to a description of the properties of the unidirectional composite material, it does not allow us to solve the problem of the prediction of these properties. These considerations therefore show the limitation of semi-empirical expressions of type (9.95), for an application to the design of composite structures.
9.5 NUMERICAL VALUES OF THE MODULI 9.5.1 Experimental Values of Moduli

Table 9.2 gives the mechanical characteristics measured on unidirectional epoxy composites, with various fibres: E- and R-glass, HM- and HS-carbon fibres, Kevlar fibres. The fibre volume fraction is the same for the three composites: 0.60. In the next subsection, we compare these experimental values with the theoretical values derived by the various relations previously established. To calculate these numerical values, it is necessary first to determine the values of the various moduli of the fibres and matrix, using the expressions: for the fibres: shear modulus bulk modulus lateral compression modulus for the matrix:
Gm = Em , 2 (1 + m ) km = Em , 3 (1 2 m ) Gf = kf = Ef , 2 (1 + f ) Ef , 3 (1 2 f )
(9.97)
K f = kf +
Gf , 3 Gm . 3 (9.98)
K m = km +
158
TABLE 9.2. Mechanical characteristics measured on various unidirectional fibre epoxy composites.
Verre E R Properties measured on monofilaments Density Youngs modulus Poisson ratio Fracture stress Ultimate elongation Characteristics measured on a unidirectional fibre-epoxy composite Fibre volume fraction Density Longitudinal Youngs modulus Transverse Youngs modulus Longitudinal shear modulus Poisson ratio Tensile fracture stress Carbone HM HR 1950 1750 380 260 0.6 1.0 2200 2500 Kevlar 49 1450 135 3500 2.5 0.60 1370 84 5.6 2.1 0.34 1400 280 70 61 1022
f (kg/m3)
Ef vf (GPa)
2600 2550 73 86 0.22 0.22 3400 4400 4.5 5.2
fu (MPa) fu (%)
Vf
0.60 0.60 2040 2010 46 52
0.60 0.60 1650 1550 230 159 14.4 14.3 4.9 4.8
c (kg/m3)
EL ET (GPa) (GPa)
10 13.6 4.6 4.7
GLT (GPa) vLT
0.31 0.31 1400 1900 1500 1500 910 970 70 70
0.32 0.32 800 1380 1250 1850 900 1430 70 80
cu (MPa)
Flexural fracture stress (MPa) Compression fracture stress (MPa) Shear fracture stress (MPa) Specific characteristics EL/c (MN m/kg)
22.5 25.9 686 945
139 103 485 890
cu /c (kN m/kg)
Epoxy matrix: Em = 3.45 GPa, m = 0.30, mu = 70 MPa, m = 1200 kg/m3.
The values obtained are reported in Table 9.3. In order to determine the moduli of the fibres, it is necessary to know the Poisson ratio f. Its value cannot be determined on the fibres directly. When this value is reported in Table 9.3 (the case of E- and R-fibres), it corresponds to a value determined on the bulk material. To find the value of Poisson ratio, when the experimental determination on the bulk material cannot be carried out, it is usual to use the value measured for LT for the unidirectional composite material (Table 9.2). The value of f is deduced from it by using the law of mixtures (9.90). It is this approach that we have used for determining (Table 9.3) the values of f for carbon and Kevlar fibres.
159
TABLEAU 9.3. Elasticity moduli of fibres and matrix.

Moduli Fibres Ef experimental (GPa) Glass E 73 0.22 (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) 29.9 43.5 53.4 R 86 0.22 35.2 51.2 62.9 Carbon HM HS 380 0.33 142.9 372.5 420.2 260 0.33 97.7 254.9 287.5 Kevlar 49 135 0.37 49.3 173.1 189.5
f experimental f calculated
Gf kf Kf
Matrix
Em experimental
3.45 0.30 1.33 2.875 3.32
m experimental
Gm km Km
9.5.2 Comparison of Experimental and Calculated Values of Moduli

Considering the values reported in Table 9.3, we have calculated the values of the moduli of the unidirectional composites, determined from the theoretical expressions considered previously. These values are reported in Table 9.4: the longitudinal Youngs modulus EL, deduced from the law of mixtures (9.82) ; the longitudinal shear modulus GLT, deduced from Relation (9.75) ; the lateral compression modulus KL, Relation (9.76) ; the transverse shear modulus GTT , Relation (9.77) ; the transverse Youngs modulus ET, derived from the preceding values of KL, GTT , LT and EL, by Relation (9.24) ; the transverse Youngs modulus ET and the longitudinal shear modulus GLT, using the simplified approaches, Relations (9.88) and (9.94) ; the lower (denoted: ) and upper (denoted: +) bounds of KL, GTT , GLT, given by Relations (9.40), (9.41) and (9.42). It is to be recalled that the lower bounds (9.40), (9.41) and (9.42) are similar to Expressions (9.76), (9.77) and (9.75).
160
TABLE 9.4. Values of the moduli of unidirectional composites, derived from the theoretical expressions.
Glass E R Exact solutions of particular problems EL (mixtures) (GPa) 45.2 53.0 0,25 0,25 0.31 4.57 8.98 3.96 10.8 0.31 4.67 9.15 4,02 11.0 Carbon HM HS 229.4 0.32 5.14 10.10 4.32 12.0 8.50 3.28 157.4 0.32 5.06 10.02 4.27 11.9 8.45 3.26 Kevlar 49 82.4 0.34 4.83 9.88 4.13 11.5 8.30 3.19 9.88 50 4.13 19.7 4.83 22.2 11.5 52
LT (mixtures) LT experimental
GLT (9.75) KL ET (9.76) (9.24) (9.88)
GTT (9.77)
(GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) (GPa)
Simplified approaches Bounds (9.40) to (9.44)
ET
8.05 8.13 3.11 3.14 8.98 9.15 22.1 25.4 3.96 4.02 11.2 12.9 4.57 4.67 13.9 16.2 10.8 11.0 28.5 32.9
GLT (9.75) KL (= KL) KL+

GTT ( = GTT ) GTT +
10.10 10.02 120 84 4.32 4.27 52.6 36.4 5.14 5.06 62.3 43 12.0 11.9 137 95
GLT (= GLT) GLT+ ET (= ET) ET+
The numerical values obtained (Table 9.4), compared with the experimental values (Table 9.2) show that: the longitudinal Youngs modulus EL is well described by the law of mixtures, for the different composites; the longitudinal shear modulus GLT is well estimated by Relation (9.75) (or lower bound (9.42)); the transverse Youngs modulus ET is quite well approximated by Relation (9.24) associated to Expressions (9.76), (9.77) and (9.90), for glass and carbon fibre composites, although the calculated values underestimate the experimental values when the Youngs modulus of the fibres increases (Rglass, HM- and HS-carbon). In contrast, the experimental values, obtained for the longitudinal shear modulus and the transverse Youngs modulus of Kevlar fibre, are very clearly lower than the theoretical values. These low values can be attributed to:
Exercises
161
a poor adhesion between aramid fibres and resins, poor properties of aramid fibres under compression and shear des fibres, a superposition of both processes. The theoretical values found for the transverse Youngs modulus and the longitudinal shear modulus by the simplified approaches yield values that are too small. This is the consequence of too important a role being accorded to the matrix in these simplified approaches. Last, the values obtained for the upper bounds show that the corresponding theoretical scheme is too removed from reality. It would fit, in actuality, a description of a composite material constituted of epoxy fibres in a glass, carbon or Kevlar matrix.
9.5.3 Conclusions
As a consequence of the preceding results, we retain the following theoretical expressions for the moduli of a unidirectional composite material: longitudinal Youngs modulus EL : law of mixtures (9.82) ; longitudinal Poisson ratio LT : law of mixtures (9.90) ; longitudinal shear modulus GLT : Relation (9.75) ; lateral compression modulus KL : Relation (9.76) ; transverse shear modulus GTT' : Relation (9.77) ; transverse Youngs modulus ET, derived from the preceding moduli using Expression (9.24). These expressions allow us to determine the variations of the moduli as functions of the volume fraction of the fibres. The corresponding curves are plotted in Figure 9.19 for E-glass fibre composites and in Figure 9.20 for HScarbon fibre composites.
EXERCISES
9.1 Calculate the stiffness and compliance constants: of a unidirectional glass fibre composite:
EL = 45 GPa, ET = 10 GPa, LT = 0.31, GLT = 4.5 GPa, GTT = 4 GPa,
of a unidirectional carbon fibre composite:

EL = 230 GPa, ET = 15 GPa, LT = 0.36, GLT = 5 GPa, GTT = 4 GPa.
9.2 Plot the moduli EL, ET, GLT and GTT' as functions of the fibre volume fraction of unidirectional composites constituted of a matrix (with the characteristics Em = 3 GPa and m = 0.30) and R-glass fibres (Ef = 86 GPa, f = 0.22).
162
80 70
60
Lateral compression modulus KL ( GPa )

1.0
Young's Moduli EL ET ( GPa)
50
60 50 40 30 20 10 0 0.0
40
EL ET
30
20
10
0.2
0.4
0.6
0.8
0 0.0
0.2
0.4
0.6
0.8
1.0

35 30
Shear moduli GLT GTT' ( GPa )
25 20 15 10 5 0
GLT GTT'
0.0
0.2
0.4
0.6
0.8
1.0
FIGURE 9.19. Variation of the different elasticity moduli of a unidirectional E-glass fibre as function of the volume fraction of the fibres.
163
300
300
Young's Moduli EL ET ( GPa)
250
Lateral compression modulus KL ( GPa )

1.0
250
200
200
150
EL
150
100
100
50
ET
0.2 0.4 0.6 0.8
50
0 0.0
0 0.0
0.2
0.4
0.6
0.8
1.0

100
Shear moduli GLT GTT' ( GPa )
80
60
40
20
GLT GTT'
0.2 0.4 0.6 0.8 1.0
0 0.0
FIGURE 9.20. Variation of the different elasticity moduli of a unidirectional HS-carbon fibre as function of the volume fraction of the fibres.
CHAPITRE 10
Elastic Behaviour of an Orthotropic Composite
10.1 HOOKES LAW FOR AN ORTHOTROPIC COMPOSITE 10.1.1 Orthotropic Composite

Laminates consist of layers of unidirectional composite materials or composites reinforced with woven fabric. Usually, the woven fabrics (Chapter 2) are made of unidirectional filaments interlaced at 90, one in the warp direction and the other in the weft direction. These layers have tree mutually orthogonal symmetry planes, and from the elastic viewpoint they behave like an orthotropic material. The material directions (1, 2) will be respectively taken in the warp and weft directions. These directions will be also denoted L and T (Figure 10.1). The direction 3 orthogonal to the plane of the layer will be also denoted T'.
3, T
2, T
weft direction
1, L
warp direction
FIGURE 10.1. Layer of orthotropic composite material.
10.1 Hookes Law for an Orthotropic Composite
165

The elastic behaviour of an orthotropic composite material is described by introducing either the stiffness constants Cij or the compliance constants Sij. Taking account the results of Chapter 7 (Relation 7.14), Hookes law is written in one of the matrix forms:
1 C11 C12 C 2 12 C22 3 C13 C23 = 0 4 0 5 0 0 0 6 0
C13 0 C23 0 C33 0 0 C44 0 0 0 0
0 0 1 0 0 2 0 0 3 , 0 0 4 C55 0 5 0 C66 6 0 1 0 2 0 3 . 0 4 0 5 S66 6
(10.1)
or
1 S11 S 2 12 3 S13 = 4 0 5 0 6 0
S12 S22 S23 0 0 0
S13 S23 S33 0 0 0
0 0 0 S44 0 0
0 0 0 0 S55 0
(10.2)
The elastic behaviour of an orthotropic composite material is thus characterized by 9 independent coefficients: C11, C12, C13, C22, C23, C33, C44, C55, C66, or S11, S12, S13, S22, S23, S33, S44, S55, S66. The stiffness and compliance matrices being the inverses of each other, we have the relations:
C11 = C22
2 S 22 S33 S 23 , S
C12 = C13 = C23 = 1 , S55 C66
S13 S 23 S12 S33 , S S12 S 23 S13 S22 , S
2 S33 S11 S13 , = S 2 S11S 22 S12
C33 = C44 =
S 1 , S 44
C55 =
S12 S13 S 23 S11 , S 1 , = S66
(10.3)
with
2 2 2 S = S11S 22 S33 S11S23 S22 S13 S33 S12 + 2 S12 S23 S13 .
166
Chapter 10 Elastic Behaviour of an Orthotropic Composite
The inverted relations giving the compliance constants as functions of the stiffness constants are obtained by interchanging the roles of Cij and Sij. A unidirectional composite is a particular case of an orthotropic composite called transversely isotropic, for which:
C13 = C12 , C33 = C22 , C44 =
1 2
( C22 C23 ) ,
C55 = C66 ,
(10.4)
and S13 = S12 , S33 = S22 ,
S44 = 2 ( S22 S23 ) , S55 = S66 .
(10.5)
10.2 ENGINEERING CONSTANTS

The usual engineering constants (Youngs moduli, Poisson ratios, shear moduli) are expressed simply as functions of compliance constants. These relations are considered hereafter.
10.2.1 Tensile Test in the Warp Direction

In the case of a tension in the warp direction, all the stresses are zero, except for the stress 1 :
1 0, i = 0
if i = 2, 3, . . . , 6.
(10.6)
As functions of the compliance constants, the elasticity equations are written as:
1 = S11 1, 2 = S12 1 , 3 = S13 1, 4 = 5 = 6 = 0.

So:
(10.7)
1 =
1 1 , S11
2 =
S12 1, S11
3 =
S13 1. S11
(10.8)
From this we deduce the Youngs modulus and the Poisson ratios, measured in a tensile test in the warp direction:
Ewp = EL = E1 = 1 , S11 S = 13 = 13 . S11
LT = 12
S = 12 , S11
(10.9)
LT
167
10.2.2 Tensile Test in the Weft Direction

In the case of a tension in the weft direction, only the stress 2 is not zero:
2 0, i = 0
The elasticity equations are written as:
if i 2.
(10.10)
1 = S12 2 , 2 = S22 2 , 3 = S23 2 , 4 = 5 = 6 = 0.

Thus: (10.11)
2 =
1 2 , S22
1 =
S12 2, S22
3 =
S23 2. S22
(10.12)
Whence the Youngs modulus and the Poisson ratios, measured in a tensile test in the weft direction:
Ewf = ET = E2 = 1 , S22 S = 23 = 23 . S22
TL
S = 21 = 12 , S 22
(10.13)
TT
10.2.3 Transverse Tension

In transverse tension, the orthotropic layer is loaded in the 3-direction normal to the (1, 2) plane of the layer:
3 0, i = 0
if i 3.
(10.14)
We easily obtained the transverse Youngs modulus and the corresponding Poisson ratios:
ET = E3 = 1 , S33 S = 23 . S33
T L
S = 31 = 13 , S33
(10.15)
T T = 32
168
10.2.4
Relations between Youngs moduli and Poisson ratios
Comparison of Relations (10.9), (10.13) and (10.15) previously established allows us to write:
LT
EL
TL
ET
LT
EL
T L
ET
TT
ET
T T
ET
(10.16)
These relations may be written in the condensed form:
ij
Ei
Ej
ji
i, j = 1, 2, 3 or L, T , T .
(10.17)
10.2.5 Shear Tests

A shear test in the plane of the orthotropic layer corresponds to a stress state such as:
6 0, i = 0
It results:
if i 6.
(10.18)
1 = 2 = 3 = 4 = 5 = 0, 6 = C66 6 .
From this we deduce the shear modulus in the layer plane as: G12 = GLT = C66 = 1 . S66
(10.19)
(10.20)
Similarly, we find the shear moduli in transverse shear tests: in the warp direction: G13 = GLT = C55 = in the weft direction: G23 = GTT = C44 = 1 . S44 (10.22) 1 , S55 (10.21)
10.3 Stiffness and Compliance as Functions of Engineering Constants
169
10.2.6 Conclusion
The elasticity relation (10.2) may be written, on introducing the engineering constants, in the form: 1 0 0 0 12 13 E E1 E1 1 1 23 12 0 0 0 1 1 E1 E2 E2 1 2 2 13 23 0 0 0 3 E1 E2 E3 3 . (10.23) = 1 4 4 0 0 0 0 0 5 G23 5 1 6 6 0 0 0 0 0 G13 1 0 0 0 0 0 G12
The elastic behaviour of an orthotropic material can be described by 9 independent constants: 3 Youngs moduli: E1 , E2 , E3 or EL , ET , ET , 3 Poisson ratios: 12 , 13 , 23 or LT , LT , TT , 3 shear moduli: G12 , G13 , G23 or GLT , GLT , GTT . The other three Poisson ratios are determined by means of Relation (10.17). (10.24)
10.3 STIFFNESS AND COMPLIANCE CONSTANTS AS FUNCTIONS OF ENGINEERING CONSTANTS 10.3.1 Compliance Constants
The expressions of stiffness constants are obtained without difficulty by considering the expressions established in the preceding section, thus:
S11 = S 22 S 44
1 , E1 1 = , E2 1 = , G23
S12 = S23 = S55
12
E1
, ,
S13 = S33 = S66
13
E1
, (10.25)
23
E2 1 = , G13
1 , E3 1 = . G12
170
10.3.2 Stiffness Constants

The expressions of the stiffness constants as functions of the engineering constants are derived from Relations (10.3) and (10.25), which yield:
C11 = C13 =
1 23 32 , E2 E3
31 + 21 32 E2 E3 + C13 = 32 12 31 E1E3
C44 = G23 , with
21 + 31 23 12 + 3213 = , E2 E3 E1E3 1 13 31 + = 13 12 23 , C22 = , E1E2 E1E3 1 12 21 + = 23 21 13 , , C33 = E1E2 E1E2

C12 = C55 = G13 , C66 = G12 ,
(10.26)
1 12 21 23 32 3113 2 21 3213 . E1E2 E3
10.3.3 Restrictions upon the Elasticity Constants

If a single stress is applied along a material direction, the strain in this direction has the same sign as the stress. From this it results that:
S11 , S 22 , S33 , S44 , S55 , S66 > 0,
(10.27)
or in terms of the engineering constants:

E1 , E2 , E3 , G23 , G13 , G12 > 0.
(10.28)
Similarly, if a single strain is applied along a material direction, the stress that results from it in this direction has the same sign as the strain applied. From this it results that:
C11 , C22 , C33 , C44 , C55 , C66 > 0,
(10.29)
and taking (10.26) into account:

1 23 32 > 0, 1 13 31 > 0, 1 12 21 > 0,
(10.30) (10.31)
and
1 12 21 23 32 3113 2 21 3213 > 0,
since the compliance matrix S is positive-definite (the determinant is positive), because the work done by all the stress components is positive. This same
Exercises
171
property associated to Relation (10.26) also implies: S23 < S22 S33 , S13 < S11S33 , S12 < S11S22 . On using the symmetry relations (10.17), the conditions (10.30) can also be written as: (10.32)
21 < 32 < 13 <
E2 , E1 E3 , E2 E1 , E3
12 < 23 < 31 <
E1 , E2 E2 , E3 E3 . E1
(10.33)
Similarly, substituting the symmetry relations into condition (10.31), we obtain:

2 2 21 3213 < 1 21
E1 2 E2 2 E3 32 13 < 1. E2 E3 E1
(10.34)
The last two conditions can be regrouped to obtain:

E E 2 E2 2 E3 >0. 21 1 + 3213 2 1 32 1 13 E3 E1 E2 E1
2
(10.35)
Finally, the preceding condition can be rearranged in such a way to obtain bounds on the Poisson ratio 21:
12 12 E2 2 E2 2 E3 3213 + 1 32 1 13 E1 E3 E1 < 21 < 12 12 E2 2 E2 2 E3 3213 1 32 1 13 E1 E3 E1
E2 E1
(10.36)
E2 . E1
EXERCISES
10.1 Calculate the stiffness and compliance constants of an orthotropic composite material with engineering constants:
172
EL = 30 GPa,
ET = 20 GPa,
ET = 10 GPa,
LT = 0.14,
GLT = 4 GPa,
LT = 0.30,
GLT = 3.5 GPa,
TT = 0.32,
GTT = 2.5 GPa.
10.2 Calculate the stiffness and compliance constants of an orthotropic composite with the characteristics:
EL = ET = 25 GPa, ET = 10 GPa,
LT = 0.12,
GLT = 4.2 GPa,
LT = 0.30,
GLT = GTT = 3.5 GPa.
TT = 0.32,
CHAPTER 11
Off-Axis Behaviour of Composite Materials
11.1 ELASTICITY RELATIONS FOR OFF-AXIS LAYERS 11.1.1 Introduction

In the preceding chapters, we have studied the elastic behaviour of a unidirectional composite material (Chapter 9) and a cloth reinforced composite (Chapter 10), expressed in the material directions: one axis in the direction of the fibres or in the warp direction, with the others being orthogonal. In fact, it was considered (Chapter 3) that laminates are processed as successive layers with fibre direction or warp direction which is shifted from one layer to the other. To analyse the elastic behaviour of such laminates, it is necessary to take a reference axis system for the whole laminate, and to refer the elastic behaviour of each layer to this reference system. Thus, a relation is needed between the stress-strain relations in the material directions of layers and those in the reference system. In this chapter, we shall thus consider (Figure 11.1) a layer of unidirectional material or cloth reinforced layer with material directions (1, 2, 3), the plane (1, 2) being identified with the plane of the layer and the direction 1 with the direction of the fibres or the warp direction. The purpose of this chapter is to characterize the elastic properties of a layer, expressing them in the reference system (1', 2', 3) of the laminate, the direction of the fibres or the warp direction making an angle with the direction 1'. This reference system is usually referred to as system (x, y, z). Hereafter, we shall indifferently use the two notations: (1', 2', 3) = (x, y, z). The first notation is better adapted to the matrix notation and is more practical for implementing changes of axis system. The second notation differentiates the material axes (1, 2, 3) = (L, T, T') from the usual Cartesian coordinate system (x, y, z) of engineers.
174
Chapter 11 Off-Axis Behaviour of Composite Materials
3, z
2, x
1, x
FIGURE 11.1. Material directions (1, 2, 3) of a laminate layer, and the reference system (1', 2', 3) = (x, y, z) of the laminate.

The elastic behaviour of a layer, referred to its material directions, is given by Relations (9.5) to (9.6) for a unidirectional layer, and (10.1) to (10.2) for a reinforced cloth layer. The stiffness C and compliance S matrices, expressed in the system (1', 2', 3), are obtained by applying the transformation equations (7.9) to (7.12) to the stiffness and compliance matrices expressed in the system (1, 2, 3). In order to apply these equations here, it is necessary to pay attention to the fact that in the present case the system change (1, 2, 3) (1', 2', 3) is performed by a rotation of angle . Thus, the relations to use are Relations (7.11) and (7.12) which here are written as: 1 C = C , (11.1) and
S = 1 S .
(11.2)
These relations, associated to Expressions (5.46), (5.48), (6.43) and (6.45), allow us to determine the stiffness matrix C and the compliance matrix S , expressed in the system (1', 2', 3). These matrices are written in the form: A11 A 12 A13 0 0 A16 A12 A22 A23 0 0 A26 A13 A23 A33 0 0 A36 0 0 0 A44 A45 0 0 0 0 A45 A55 0 A16 A26 A36 , 0 0 A66
or Sij . The expressions for the stiffness and compliance constants in with Aij = Cij the system (1', 2', 3), that are deduced from Relations (11.1) and (11.2), are reported in Tables (11.1) and (11.2) for a unidirectional composite, and in Tables (11.3) and (11.4) for an orthotropic composite.
11.1 Elasticity Relations for Off-Axis Layers
175
TABLE 11.1. Stiffness constants of a unidirectional composite, the fibres direction of which makes an angle with the reference x-direction (Figure 11.1).
= C11 cos 4 + C22 sin 4 + 2 ( C12 + 2C66 ) sin 2 cos 2 , C11 = ( C11 + C22 4C66 ) sin 2 cos 2 + C12 sin 4 + cos 4 , C12 = C12 cos 2 + C23 sin 2 , C13 = 0, C14 = 0, C15
= ( C11 C12 2C66 ) sin cos3 + ( C12 C22 + 2C66 ) sin 3 cos , C16 = C11 sin 4 + C22 cos 4 + 2 ( C12 + 2C66 ) sin 2 cos 2 , C22 = C12 sin 2 + C23 cos 2 , C23 = 0, C24 = 0, C25
= ( C11 C12 2C66 ) sin 3 cos + ( C12 C22 + 2C66 ) sin cos3 , C26 = C22 , C33 = ( C12 C23 ) sin cos , C36 = C44
1 2
= 0, C34
= 0, C35
( C22 C23 ) cos2 + C66 sin 2 ,

= 0, C46 = 0, C56
1 = C45 C66 2 ( C22 C23 ) sin cos ,
= C55
1 2
( C22 C23 ) sin 2 + C66 cos2 ,
2 2 4 4 = C66 C11 + C22 2 ( C12 + C66 ) sin cos + C66 sin + cos .
The elasticity relations, referred to the system (1', 2', 3), are thus written in one of the two forms, as function of the stiffness constants:
C12 xx C11 yy C12 C22 zz C13 C23 = 0 yz 0 xz 0 0 C26 C16 xy C13 C23 C33 0 0 C36 0 0 0 C44 C45 0 0 0 0 C45 C55 0 xx C16 C26 yy zz C36 , 0 yz 0 xz C66 xy
(11.3)
176
TABLE 11.2. Compliance constants of a unidirectional composite, off its material directions.
= S11 cos 4 + S22 sin 4 + ( 2 S12 + S66 ) sin 2 cos 2 , S11 = ( S11 + S22 S66 ) sin 2 cos 2 + S12 sin 4 + cos 4 , S12 = S12 cos 2 + S23 sin 2 , S13 = 0, S14 = 0, S15
3 3 = S16 2 ( S11 S12 ) S66 sin cos + 2 ( S12 S22 ) + S66 sin cos ,
= S11 sin 4 + S22 cos 4 + ( 2 S12 + S66 ) sin 2 cos 2 , S22 = S12 sin 2 + S23 cos 2 , S23 = 0, S24 = 0, S25
3 3 = S26 2 ( S11 S12 ) S66 sin cos + 2 ( S12 S 22 ) + S66 sin cos ,
= S22 , S33
= 0, S34
= 0, S35
= 2 ( S12 S23 ) sin cos , S36 = 2 ( S 22 S 23 ) cos 2 + S66 sin 2 , S 44 = S 45 S66 2 ( S22 S23 ) sin cos , = 2 ( S22 S 23 ) sin 2 + S66 cos 2 , S55 = 0, S46 = 0, S56
2 2 4 4 = 2 S66 2 ( S11 + S 22 2 S12 ) S66 sin cos + S66 sin + cos .
or as function of the compliance constants:

xx S11 yy S12 zz S13 = yz 0 xz 0 S16 xy S12 S22 S 23 0 0 S26 S13 S 23 S33 0 0 S36 0 0 0 S 44 S 45 0 0 0 0 S 45 S55 0 xx S16 S 26 yy zz S36 . 0 yz 0 xz S66 xy
(11.4)
177
TABLE 11.3. Stiffness constants of an orthotropic composite, the material direction 1 of which makes an angle with the reference x-direction (Figure 11.1).
= C11 cos 4 + C22 sin 4 + 2 ( C12 + 2C66 ) sin 2 cos 2 , C11 = ( C11 + C22 4C66 ) sin 2 cos 2 + C12 sin 4 + cos 4 , C12 = C13 cos 2 + C23 sin 2 , C13 = 0, C14 = 0, C15
= ( C11 C12 2C66 ) sin cos3 + ( C12 C22 + 2C66 ) sin 3 cos , C16 = C11 sin 4 + C22 cos 4 + 2 ( C12 + 2C66 ) sin 2 cos 2 , C22 = C13 sin 2 + C23 cos 2 , C23 = 0, C24 = 0, C25
= ( C11 C12 2C66 ) sin 3 cos + ( C12 C22 + 2C66 ) sin cos3 , C26 = C33 , C33 = ( C13 C23 ) sin cos , C36 = C44 cos 2 + C55 sin 2 , C44 = ( C55 C44 ) sin cos , C45 = C44 sin 2 + C55 cos 2 , C55 = 0, C46 = 0, C56 = 0, C34 = 0, C35
2 2 4 4 = C66 C11 + C22 2 ( C12 + C66 ) sin cos + C66 sin + cos .
178
TABLE 11.4. Stiffness constants of an orthotropic composite, off its material directions.
= S11 cos 4 + S22 sin 4 + ( 2S12 + S66 ) sin 2 cos 2 , S11 = ( S11 + S22 S66 ) sin 2 cos 2 + S12 sin 4 + cos 4 , S12 = S13 cos 2 + S 23 sin 2 , S13 = 0, S14 = 0, S15
3 3 = S16 2 ( S11 S12 ) S66 sin cos + 2 ( S12 S 22 ) + S66 sin cos ,
= S11 sin 4 + S22 cos 4 + ( 2 S12 + S66 ) sin 2 cos 2 , S 22 = S13 sin 2 + S23 cos 2 , S 23 = 0, S 24 = 0, S 25
3 3 = S 26 2 ( S11 S12 ) S66 sin cos + 2 ( S12 S 22 ) + S66 sin cos ,
= S33 , S33
= 0, S34
= 0, S35
= 2 ( S13 S 23 ) sin cos , S36 = S 44 cos 2 + S55 sin 2 , S 44 = ( S55 S 44 ) sin cos , S 45 = S 44 sin 2 + S55 cos 2 , S55 = 0, S46 = 0, S56
2 2 4 4 = 2 S66 2 ( S11 + S 22 2 S12 ) S66 sin cos + S66 sin + cos .
179
Comparing the expressions in Tables 11.1 to 11.4 shows that the expressions relating to a unidirectional material and to an orthotropic material are identical for or Sij with i, j = 1, 2, 6. For example: the terms Cij
identical relations C12 C16 C11 C C C 22 26 12 C26 C66 C16 C23 C36 C13 0 0 0 0 0 0 C13 C23 C36 C33 0 0
different relations 0 0 0 0 C44 C45 0 0 0 . 0 C45 C55
(11.5)
11.1.3 Other Expressions for Stiffness Matrices

The expressions for the stiffness and compliance constants off the material directions can be rewritten by introducing multiples of the angle . In fact, we have:
cos 4 = 1 3 + 4 cos 2 + cos 4 ) , 8( cos 2 sin 2 = 1 1 cos 4 ) , 8( sin 4 = 1 3 4 cos 2 + cos 4 ) , 8( sin 2 =
1 2
cos3 sin = 1 2sin 2 + sin 4 ) , 8( cos sin 3 = 1 2sin 2 sin 4 ) , 8( sin cos = 1 sin 2 , 2 cos 2 =
1 2
(1 cos 2 ) ,
(1 + cos 2 ) .
These relations allow us to write, for example:

=1 C11 3C + 3C22 + 2C12 + 4C66 ) + 1 C C22 ) cos 2 8 ( 11 2 ( 11 +1 C + C22 2C12 4C66 ) cos 4 . 8 ( 11
Thus: introducing:
= U1 + U 2 cos 2 + U 3 cos 4 , C11
U1 = 1 3C + 3C22 + 2C12 + 4C66 ) , 8 ( 11 U2 = 1 C C22 ) , 2 ( 11 U3 = 1 C + C22 2C12 4C66 ) . 8 ( 11
Similar transformations can be carried out for each constant. The results obtained in the case of a unidirectional composite are reported in Table 11.5. The interest of these relations is that they show the terms which are invariant under a rotation of the direction of the fibres. Analogous expressions can be derived for the compliance constants.
180
TABLE 11.5. Expressions for the stiffness constants of a unidiretional composite as functions of the multiples of the fibre orientation.
= U1 + U 2 cos 2 + U 3 cos 4 , C11 = U 4 U 3 cos 4 , C12 =1 C16 U sin 2 + U 3 sin 4 , 2 2 = U1 U 2 cos 2 + U 3 cos 4 , C22 = U 6 U 7 cos 2 , C23 =1 C26 U sin 2 U 3 sin 4 , 2 2 = C22 , C33 = U 8 + U 9 cos 2 , C44 = U 8 + U10 cos 2 , C55 = U 7 sin 2 , C36 = U10 sin 2 , C45 = U 5 U 3 cos 4 , C66 = U 6 + U 7 cos 2 , C13
with
3C + 3C22 + 2C12 + 4C66 ) , U1 = 1 8 ( 11
1 C C 11 22 , 2 U3 = 1 C11 + C22 2C12 4C66 8 U4 = 1 C11 + C22 + 6C12 4C66 8 U5 = 1 C11 + C22 2C12 + 4C66 8 U6 = 1 C12 + C23 , 2 U7 = 1 C12 C23 , 2 U8 = 1 C22 C23 + 2C66 , 4 U9 = 1 C22 C23 2C66 , 4 2C66 C22 + C23 . U10 = 1 4
U2 =
( ( ( ( ( ( ( ( (
), ), U U4 ) , )= 1 2( 1
) )
) )
11.2 ELASTICITY MODULI 11.2.1 Expressions for Off-Axis Moduli

The elasticity moduli or engineering constants are the practical constants that engineers use to describe the mechanical behaviour of a material. These constants
11.2 Elasticity Moduli
181
are determined in particular tests and are easily expressed as functions of the compliance constants of material.
11.2.1.1 Off-Axis Tensile Test

1. Tension in the x-direction
In the case of a tensile test in the x-direction, all stresses are zero, except for the stress xx :
= xx 0, 1
i = 0,
i = 2, . . . , 6.
Introducing the compliance constants, the elasticity relations (11.4) are written as: xx , xx = S11 xx , yy = S12 xx , zz = S13 (11.6)
yz = xz = 0,
xx . xy = S16 The Youngs modulus Ex in the x-direction is defined by: Ex =
xx 1 = . xx S11
(11.7)
Considering Table 11.2 or 11.4, we obtain: 1 = S11 cos 4 + S22 sin 4 + ( 2S12 + S66 ) sin 2 cos 2 . Ex (11.8)
This expression can be rewritten by introducing the engineering constants (Chapters 9 and 10) of the unidirectional or orthotropic composite, measured in its material directions: EL = Whence: 1 1 1 1 cos 4 + sin 4 + = 2 LT E x EL ET EL GLT 2 2 sin cos . (11.9) 1 , S11
LT =
S12 , S11
ET =
1 , S12
GLT =
1 . S66
The normal strains yy and zz in the transverse directions are related to the normal strain xx in the x-direction by Expressions (11.6), which lead to:
yy =
S12 xx , S11
zz =
S13 xx . S11
182
These relations allow us to derive the Poisson ratios xy and xz defined by:
yy = xy xx ,
Thus, we otain: S12 , S11
zz = xz xx .
S13 . S11
xy =
xz =
Whence the expressions for the Poisson ratios:

1 1 1 + xy = E x LT cos 4 + sin 4 EL ET GLT EL
2 2 sin cos , (11.10)
xz = Ex
LT cos 2 + TT sin 2 , ET EL
(11.11)
with LT' = LT for an unidirectional composite. Lastly, Expressions (11.6) show that the off-tension induces an in-plane shear strain xy. We then define a coupling coefficient xy,x, analogous to a Poisson ratio, which relates the in-plane shear strain to the normal strain xx in the xdirection by the relation: xy = xy , x xx . (11.12) And so: Ex . xy , x = S16 Considering Tables 11.2 and 11.4, the coupling coefficient is finally expressed as follows:
xy , x = Ex
1 2 3 2 LT cos sin G E E LT L L 2 1 + + 2 LT EL GLT ET 3 sin cos .
(11.13)
2. Tension in the y-direction
A tensile test in the y-direction can also be considered. However, this test does not give any new information, since this test reduces to a rotation of the tension direction through an angle equal to /2. For example, the Youngs modulus in the y-direction Ey, relating the stress yy to the strain yy, is deduced from Relation (11.8), replacing by + / 2. Thus:
1 1 1 1 = sin 4 + cos 4 + = S22 2 LT Ey EL ET EL GLT
2 2 sin cos . (11.14)
183
Similarly, we define the coupling coefficient xy,y, which relates the shear strain xy to the strain yy by the relation:
xy = xy , y yy = xy , y
The expression of this coefficient is expressed as:
yy
Ey
(11.15)
xy , y = E y
1 2 3 2 LT sin cos EL EL GLT 2 1 + + 2 LT EL GLT ET 3 cos sin .
(11.16)
11.2.1.2 In-Plane Shear Test

We consider the case of a shear test in the plane of a layer, corresponding to the stress state:
= xy 0, 6
i = 0,
i = 1, . . . , 5.
The elasticity relations (11.4) are written as:
xy , xx = S16 xy , yy = S26 xy , zz = S36

(11.17)
yz = xz = 0,
xy . xy = S66
The shear modulus Gxy in this test is defined by:
Gxy =
xy 1 = . xy S66
(11.18)
Thus, from Tables 11.2 and 11.4, we deduce: 2 1 2 1 2 2 = 2 + + 4 LT sin cos Gxy E E E G T L LT L 1 sin 4 + cos 4 . + GLT
(11.19)
184
The in-plane shear test induces normal strains xx, yy, zz respectively in the three directions (x, y, z). The first two strains are expressed by:
xx = xy , x yy = xy , y
xy
Ex
= xy , x = xy , y
Gxy Ex Gxy Ey
xy ,
(11.20)
xy
Ey
xy .
By analogy to the preceding relations, the normal strain zz in the z-direction can be written in the form:
zz = xy , z
xy
ET
= xy , z
Gxy ET
xy .
(11.21)
The coupling coefficient thus introduced is given by the expression:

= 2 ( TL TT ) sin cos . xy , z = ET S36
(11.22)
11.2.1.3 Transverse Shear Test

The elasticity moduli, introduced in the preceding tests off the material directions, involve only four of the principal moduli: EL, ET, GLT, and LT, in the case of a unidirectional composite, and five of the principal moduli: EL, ET, GLT, LT, and LT', in the case of an orthotropic material. We shall also see (Section 11.3) that the moduli EL, ET, GLT, and LT suffice to describe a two-dimensional behaviour of a composite layer. The description of more general elasticity problems requires the elasticity moduli for a transverse shear test to be known. Such a test is characterized by: = xz 0, 5 i = 0 si i = 1, 2, 3, 4, 6, or
4 = yz 0, (11.23) i 0 si 1, 2, 3, 5, 6. = = i
The two tests differ simply by the reversal of the respective roles of the x and y directions. In the first case, the elasticity relations (11.4) are written:
xx = 0,
xz , yz = S45 xz , xz = S55
yy = 0,
zz = 0,
(11.24)
xy = 0.
These relations show that the transverse shear state induces only transverse deformations xz and yz.
185
The shear modulus Gxz in this test is defined by: Gxz =
xz 1 = . xz S55
(11.25)
In the case of an orthotropic composite, we deduce the expression for Gxz from Table 11.4, hence: 1 1 1 = sin 2 + cos 2 . Gxz GTT GLT (11.26)
This relation can be rewritten in the case of a unidirectional composite, considering Expression (9.19) of the transverse shear modulus as function of the transverse Youngs modulus. Whence:
2 (1 + TT ) 2 1 1 = sin + cos 2 . Gxz ET GLT
(11.27)
The shear test also induces a shear strain yz. We thus define a coupling coefficient xz,yz, relating the shear strain yz to the shear strain xz by the relation:
yz = xz , yz xz = xz , yz
xz
Gxz
(11.28)
The expression of this coupling coefficient is obtained easily and is written as: = Gxz xz , yz = Gxz S45 1 1 sin cos . GLT GTT (11.29)
In the case where the longitudinal GLT' and transverse GTT' shear moduli are close, the coupling coefficient is practically zero, and the induced coupling is negligible. The preceding expression of the coupling coefficient can be rewritten in the case of a unidirectional material as:
xz , yz = Gxz
2 (1 + TT ) 1 sin cos . ET GLT
(11.30)
11.2.2
Variations of the Elasticity Moduli of a Unidirectional Composite
To better appreciate the variations of the elasticity moduli, we can plot their graphs as functions of the orientation of the fibres, for usual composites. We consider the three following cases of unidirectional.
186
1. Glass fibre-epoxyde composites, with:
EL = 46 GPa, GLT = 4.7 GPa,
ET = 10 GPa, GTT = 4 GPa.
LT = 0.31,
(11.31)
2. Carbon fibre-epoxyde composites, with:
EL = 159 GPa, GLT = 4.8 GPa,
ET = 14 GPa, GTT = 4.3 GPa.
LT = 0.32,
(11.32)
3. Kevlar fibre-epoxyde composites, with:
EL = 84 GPa, GLT = 2.1 GPa,
ET = 5, 6 GPa, GTT = GLT = 2.1 GPa.
LT = 0.34,
(11.33)
The variations of the moduli Ex, Gxy, xy and xy,x as functions of the orientation of fibres are reported in Figures 11.2 to 11.4. In the case of glass fibre composites, the Youngs modulus in the x-direction Ex decreases monotonically from the value EL for = 0 to the value ET for = 90. The shear modulus Gxy passes through a maximum for = 45, and its variation is symmetric about either side of this value. The Poisson ratio xy also passes through a maximum for a value of the angle that depends upon the composite. The coupling coefficient xy,x is zero for = 0 and = 90, and reaches high values for intermediate angle values. The curves also show that the extremum values of Gxy, xy, xy,x are reached for fibre orientations different from the material directions. This property is also observed for the Youngs modulus Ex (Figures 11.3 and 11.4), for composites with carbon and Kevlar fibres. In fact, searching for the extremal values of Ex from Equation (11.9), we easily find that the modulus Ex passes through a maximum greater than EL for a value of different from 0, if:
GLT > EL , 2 (1 + LT )
(11.34)
and that the modulus Ex passes through a minimum lower than ET for a value of different from 90, if: GLT < EL E 2 L + LT ET . (11.35)
This latter equality is satisfied in the case of carbon fibre-epoxyde composites (11.32) as well as in the case of Kevlar fibre-epoxyde composites (11.33).
187
50
8 7
Young's modulus Ex (GPa)
40
Shear modulus Gxy (GPa)

0 15 30 45 60 75 90
6 5 4 3 2 1
30
20
10
15
30
45
60
75
90
Orientation ()
1.2
Orientation ()
Poisson ratio and Coupling coefficient
1.0
0.8
xy,x
0.6
0.4
xy
0.2
0.0
15
30
45
60
75
90
Orientation ()
FIGURE 11.2. Variations of the engineering constants in the case of a glass fibre-epoxide composite.
188
160 140
14
12
120 100 80 60 40 20 0
10
15
30
45
60
75
90
15
30
45
60
75
90
Orientation ()
3.0
Orientation ()
2.5
2.0
1.5
xy,x
1.0
xy
0.5 0.0
-0.5
15
30
45
60
75
90
Orientation ()
FIGURE 11.3. Variations of the engineering constants in the case of a carbon fibreepoxide composite.
189
90 80
60 50 40 30 20 10 0

0 15 30 45 60 75 90
70
15
30
45
60
75
90
Orientation ()
3.0
Orientation ()
2.5
2.0
1.5
xy,x
1.0
xy
0.5 0.0
-0.5
15
30
45
60
75
90
Orientation ()
FIGURE 11.3. Variations of the engineering constants in the case of a Kevlar fibreepoxide composite.
190
11.3 PLANE STRESS STATE 11.3.1 Introduction

The basic elements developed in the preceding sections can be applied to the solution of any arbitrary elasticity problem for a composite material. In the case where the elasticity problem can be reduced to a two-dimensional problem, the relations established previously in the general case simplify. In this section, we consider the case of a two-dimensional stress state the results of which are necessary in the study of the mechanical behaviour of laminates which will be developed in Part 4.
11.3.2 Two-Dimensional Stress State

A two-dimensional stress state is characterized by a stress tensor of the following form:
xx xy ( M ) = xy yy 0 0 0 0 , 0
(11.36)
at each point M of the material. The z-direction is a principal direction with a zero eigenvalue. Usually such a stress state is called a plane stress state. In fact, it is necessary to distinguish between an effective plane stress state and a twodimensional stress state. In the strict sense, a plane stress state is a particular case of two-dimensional stresses for which the stress tensor components are independent of the z coordinate.
11.3.3
Elasticity Equations for Plane Stress State
The stresses (11.36) at point M can be written in the form:

11 xx 1 2 22 yy 0 0 0 . = = 0 0 0 0 0 0 6 12 xy
11.3 Plane Stress State
191
Hence:
i 0 i = 0
S11 1 S 2 12 S13 3 = 0 4 5 0 6 S16 S12 S22 S23 0 0 S26
if if
i = 1, 2, 6, i = 3, 4, 5.
(11.37)
The strains are obtained by the expression: S13 S23 S33 0 0 S36 0 0 0 S44 S45 0 0 0 0 S45 S55 0 1 S16 2 S26 0 S36 . 0 0 0 0 S66 6
Whence:
= S11 1 + S12 2 + S16 6 , 1 = S12 1 + S 22 2 + S26 6 , 2 = S13 1 + S23 2 + S36 6 , 3 = 0, 4 = 0, 5 = S16 1 + S26 2 + S66 6 . 6 From this we deduce that: (11.38)
i 0 i = 0
if if
i = 1, 2, 3, 6, i = 4, 5.
(11.39)
Thus, there exists a strain in the z-direction, given by: = S13 1 + S23 2 + S36 6 zz = 3 xx + S23 yy + S36 xy . = S13 (11.40)
The first two terms are coupling terms induced by the Poisson effect, and the third term results from a shear coupling. The relations between stresses and strains given in terms of stiffness constants are expressed as: C11 C12 C13 0 1 C C C 0 22 23 2 12 C23 C33 0 C13 0 = 0 0 C44 0 0 0 0 0 0 C45 C26 C36 0 6 C16 1 0 C16 0 C26 2 3 0 C36 . 0 0 C45 0 0 C55 0 C66 6
192
Thus: = C11 1 + C12 2 + C13 3 + C16 6 , 1 = C12 1 + C22 2 + C23 3 + C26 6 , 2 1 + C23 2 + C33 3 + C36 6 , 0 = C13 = C16 1 + C26 2 + C36 3 + C66 6 . 6
, 2 , 3 and 6 are not independent. We These relations show that the strains 1 deduce that:
(11.41)
= 3
1 1 + C23 2 + C36 6 ). ( C13 C33
(11.42)
This expression, associated with Relations (11.41), then allows to express the , 2 and 6 as functions of the strains 1 , 2 and 6 . For example, we stresses 1 have:
= C11 1
2 C13 C C C C + C12 13 23 2 + C16 13 36 6 , 1 C33 C C 33 33
and 6 . These three relations in 1 , 2 and 6 and analogous expressions for 2 can then be put in the matrix form: Q11 Q12 1 = Q Q 22 2 12 6 Q16 Q26 1 Q16 2 Q26 , Q66 6
(11.43)
introducing:
= Cij Qij
Ci3C j 3 , C33
i, j = 1, 2, 6,
with
(11.44)
. Qji = Qij are called the reduced stiffness constants in a plane stress The coefficients Qij state. The matrix:
Q12 Q11 Q22 Q = Q12 Q26 Q16 Q16 Q26 Q66
(11.45)
is the reduced stiffness matrix.
193
In conclusion, in the case of a plane stress state, we have:
i 0 if i = 1, 2, 6 i = 0 if i = 3, 4, 5,
and
i 0 if i = 1, 2, 3, 6 i = 0 if i = 4, 5.
The elasticity relations can be written in one of the two forms:

S11 1 = S 2 12 6 S16 S12 S 22 S 26 1 S16 2 S 26 S66 6
(11.46)
with or
= S13 1 + S 23 2 + S36 6 , 3 Q11 Q12 1 = Q Q 22 2 12 Q26 6 Q16 1 Q16 Q26 2 Q66 6
with = 3 1 1 + C23 2 + C36 6 ). ( C13 C33
The reduced stiffness constants are expressed as functions of the stiffness constants by the relations:
= Cij Qij Ci3C j 3 , C33 i, j = 1, 2, 6,
. Qji = Qij
The matrices Sij and Qij are inverses of each other.
11.3.4 Reduced Stiffness Matrix in Material Directions

In the material directions of an orthotropic composite, the stiffness constants are such that:
= C11, C11 = C22 , C22 = C66 , C66 = C12 , C12 = C23 , C23 = C26 = C36 = 0. C16 = C13 , C13 = C33 , C33
194
Considering these relations, Expressions (11.44) allow to find the reduced stiffness constants expressed in the material directions, whence:
Q11 = C11 Q22
2 C13 , C33
Q12 = C12 Q26 = 0,
C13C23 , Q16 = 0, C33 Q66 = C66 ,
2 C23 , = C22 C33
(11.47)
with, in addition, for a unidirectional composite:

C13 = C12 , C33 = C22 .
The reduced stiffness matrix, expressed in the material directions, is thus written:
Q11 Q12 Q= Q12 Q22 0 0
0 . 0 Q66
(11.48)
When referred to the material directions, a plane stress state is characterized by: i 0 if i = 1, 2, 6 i 0 if i = 1, 2, 3, 6 and i = 0 if i = 3, 4, 5, i = 0 if i = 4, 5. The elasticity relations are written in one of the two forms:
1 S11 = S 2 12 6 0 S12 S 22
0 0 1 0 2 S66 6
(11.49)
with or
3 = S13 1 + S23 2 ,
1 Q11 Q12 = Q 2 12 Q22 0 6 0
0 1 0 2 Q66 6
(11.50)
with
3 =
1 ( C131 + C23 2 ) . C33
The reduced stiffness constants are expressed by Relations (11.47). The matrices Sij and Qij are inverses of each other.
195
Because the matrices Qij are inverses, we deduce the relations: Sij and
Q11 = Q22 = S22
2 S11S22 S12
, ,
Q12 = Q66
S12
S11
2 S11S 22 S12
2 S11S 22 S12 1 . = S66
, (11.51)
The preceding relations allow us to express the reduced stiffness constants as functions of the engineering constants in the material directions (Chapters 9 and 10). We obtain:
Q11 = EL = EL
2 1 LT
1 LT TL ET
ET EL ET EL
Q22 =
1 LT TL
ET
2 1 LT
ET Q11 , EL
(11.52)
Q12 =
LT ET = LT Q22 , 1 LT TL
Q66 = GLT .
11.3.5
Relations between the Off-Axis and Material Axes Reduced Stiffness Constants
The relations between the off-axis reduced stiffness constants and those expressed in the material directions only bring in the 1, 2 and 6 components of the stresses and strains. The structure of the general relations (9.5) for a unidirectional composite and (10.1) for an orthotropic composite, of the stress-strain relations (11.43) in a two-dimensional stress state, of the transformation equations (Tables 11.1 and 11.3) established in the general case, shows that the transformation as functions of equations which express the reduced stiffness constants Qij constants Qij are identical to those obtained for the stiffness constants Cij (Tables 11.1 and 11.3), when they are restricted to i, j = 1, 2, 6. Lastly, these elements associated with the remark made in Subsection 11.1.2 (Relation (11.5)) show that these equations are also identical for a unidirectional composite as well as for an orthotropic composite. The transposed results of Tables (11.1) and (11.3) are reported in Table 11.6. In the same way as in the general case (Table 11.5), it is possible to rewrite the transformation equations by introducing multiples of angle . These expressions are reported in Table 11.7.
196
TABLE 11.6. Reduced Stiffness constants of a unidirectional or orthotropic composite, off its material directions (Figure 11.1).
= Q11 cos 4 + Q22 sin 4 + 2 ( Q12 + 2Q66 ) sin 2 cos 2 , Q11 = ( Q11 + Q22 4Q66 ) sin 2 cos 2 + Q12 sin 4 + cos 4 , Q12 = ( Q11 Q12 2Q66 ) sin cos3 + ( Q12 Q22 + 2Q66 ) sin 3 cos , Q16 = Q11 sin 4 + Q22 cos 4 + 2 ( Q12 + 2Q66 ) sin 2 cos 2 , Q22 = ( Q11 Q12 2Q66 ) sin 3 cos + ( Q12 Q22 + 2Q66 ) sin cos3 , Q26
2 2 4 4 = Q66 Q11 + Q22 2 ( Q12 + Q66 ) sin cos + Q66 sin + cos .
TABLEAU 11.7. Expressions for reduced stiffness constants as function of multiple angles of orientation.
= V1 + V2 cos 2 + V3 cos 4 , Q11 = V4 V3 cos 4 , Q12 =1 Q16 V sin 2 + V3 sin 4 , 2 2 = V1 V2 cos 2 + V3 cos 4 , Q22 =1 Q26 V sin 2 V3 sin 4 , 2 2 = V5 V3 cos 4 , Q66 with V1 = 1 3Q + 3Q22 + 2Q12 + 4Q66 ) , 8 ( 11 V2 =
1 2
( Q11 Q22 ) ,
V3 = 1 Q + Q22 2Q12 4Q66 ) , 8 ( 11 V4 = 1 Q + Q22 + 6Q12 4Q66 ) , 8 ( 11 V5 = 1 Q + Q22 2Q12 + 4Q66 ) = 8 ( 11

1 2
(V1 V4 ) .
197
11.3.6 Conclusions
Any plane stress (xx, yy, xy) of a laminate layer is characterized (11.43) by , referred to the reference axes (x, y, z) of the the reduced stiffness constants Qij laminate. These constants are given (Tables 11.6 and 11.7) as functions of the reduced stiffness constants Qij, referred to the material directions of the layer considered. These expressions are identical for a unidirectional as well as for an orthotropic layer. The stiffness constants Qij (Q11, Q12, Q22, Q66) are themselves expressed (11.52) as functions of the engineering constants EL, ET, LT, GLT (or E1, E2, 12, G12), measured in the material directions (1, 2, 3) = (L, T, T') of the layer: for a unidirectional layer, L is the direction of fibres; for an orthotropic layer, L is the warp direction and T the weft direction.
11.3.7 Example
A unidirectional layer is subjected in its (x, y) plane (Figure 11.5) to the following strain state:
xx = 1 % = 102 , yy = 0,5 % = 5 103 , xy = 2 % = 2 102.

The direction of the fibres makes an angle of 30 with the x-direction. The engineering constants of the composite material are:
EL = 40 GPa, ET = 10 GPa,
LT = 0.32,
GLT = 4.5 GPa.
y L
xy
yy
xx
30 x
FIGURE 11.5. Example of an application.
198
By considering that the layer is in a plane stress state, determine: 1. the stresses xx, yy, xy in the direction system (x, y); 2. the stresses in the material directions (L, T) of the layer.
1. Determination of the Stresses xx, yy and xy
We have to determine first the reduced stiffness matrix referred to the material directions: Q11 = EL
2 1 LT
ET EL
= 41.051 GPa,
Q 22 =
ET Q11 = 10.263 GPa, EL
Q12 = LT Q22 = 3.284 GPa, Q66 = GLT = 4.5 GPa. Whence the reduced stiffness matrix expressed in the material axes is:
0 41, 051 3, 284 Q = 3, 284 10, 263 0 GPa . 0 4,5 0
The reduced stiffness matrix, referred to the axes (x, y), is next derived from the expressions in Table 11.6: 9 1 31 + 10.263 + 2 ( 3.284 + 2 4.5 ) = 28.339 GPa, 16 16 44 31 9 1 = ( 41.051 + 10.263 4 4.5 ) Q12 + 3.284 + = 8.299 GPa, 44 16 16 = 41.051 Q11 = ( 41.051 3.284 2 4.5 ) Q16 13 3 1 3 + ( 3.284 10.263 + 2 4.5 ) 2 8 8 2
= 9.561 GPa, 31 1 9 = 41.051 + 10.263 + 2 ( 3.284 + 2 4.5 ) Q22 = 12.945 GPa, 16 44 16 1 3 3 31 = ( 41.051 3.284 2 4.5 ) Q26 + ( 3.284 10.263 + 2 4.5 ) 8 2 8 2 = 3.770 GPa, 13 9 1 = Q66 41.051 + 10.263 2 ( 3.284 + 4.5 ) 4 4 + 4.5 16 + 16 = 9.515 GPa.
199
Whence the reduced stiffness matrix in the (x, y) axes is:

28.339 8.299 9.561 . Q = 8.299 12.945 3.770 GPa 9.561 3.770 9.515
The stresses in the (x, y) axes are next deduced from (11.43) as:
xx 28.339 8.299 9.561 10 9 3 yy = 8.299 12.945 3.770 10 5 10 . xy 20 9.561 3.770 9.515 Thus:
xx = 433 MPa, yy = 94 MPa, xy = 267 MPa.

2. Determination of the Stresses in the Material Directions
The stresses in the material axes are obtained from the general equation (5.44). In the case of plane stresses, this relation is restricted to the three stresses in the plane and is written as:
2 L cos = sin 2 T LT sin cos
sin 2 cos 2 sin cos
2sin cos xx 2sin cos yy , xy cos 2 sin 2
(11.53)
where is the angle of the fibre direction with the reference x-direction. In the present case, this expression is written:
3 L 4 = 1 T 4 LT 1 3 4 Thus: 3 433 3 1 94 2 3 4 MPa . 1 3 1 267 4 2
1 4 1 2
L = 580 MPa, T = 53 MPa, LT = 13.5 MPa.

These results are illustrated in Figure 11.6.
200
T
T = 53 MPa
L
L = 580 MPa LT = 13.5 MPa
FIGURE 11.6. Stresses in the material directions.
11.4 EXPERIMENTAL DETERMINATION OF ENGINEERING CONSTANTS 11.4.1 Introduction

The elastic behaviour of a composite material is completely determined by the knowledge of five coefficients in the case of a unidirectional composite, and of nine coefficients in the case of an orthotropic material. In the case of a plane stress state, only four coefficients are necessary: EL, ET, LT, GLT, whether the materials are unidirectional or orthotropic. These moduli can be measured by implementing tensile tests and in-plane shear tests.
11.4.2 Longitudinal Tensile Test

In a longitudinal tensile test, a load F1 is applied in the direction of the fibres (for a unidirectional composite) or in the warp direction (for a cloth reinforced composite) on the cross section S1 of the test specimen (Figure 11.7). The stress 11 is given by: F 11 = 1 . (11.54) S1 The experimental procedure consists in measuring in the gauge length of the test specimen: the load F1, the elongation l1 of the longitudinal length l1, the variation l2 of the transverse dimension l2.
11.4 Experimental Determination of Engineering Constants
201
F1
l2
F1
l1
FIGURE 11.7. Longitudinal tensile test.
The longitudinal and transverse strains are respectively given by:
11 =
l1 l1
and
22 =
l2 . l2
(11.55)
The longitudinal modulus EL and the Poisson ratio LT are next calculated from the expressions:
EL =
11 11
and
LT =
22 . 11
(11.56)
11.4.3 Transverse Tensile Test

In a transverse tensile test, a load F2 is applied in the transverse direction to the fibres or in the weft direction (Figure 11.8). The stress 22 is then given by:
22 =
F2 , S2
(11.57)
where S2 is the area of the cross section to which the load F2 is applied. As in the longitudinal tensile test, the strains 11 and 22 are measured in the gauge part of the test specimen. The transverse modulus ET and the Poisson ratio TL are next calculated by the relations:
ET =
22 22
and
TL =
11 . 22
(11.58)
202
F2
l1
F2
l2
Figure 11.8. Transverse tensile test.
At this stage, the moduli measurement must satisfied, to within experimental errors, Relation (10.16). If this relation is not satisfied, two reasons can be invoked: 1. the experimental values have not been measured correctly; 2. the material cannot be described by a linear stress-strain relation.
11.4.4 Off-Axis Tension

A test usually considered for measuring the in-plane shear modulus GLT is a tensile test implemented at 45 to the direction of the fibres or the weft direction (Figure 11.9). The load F is applied in the x-direction on the cross-section S, involving a normal stress: F 45 = . (11.59) S
l1
F
45
l2
L
FIGURE 11.9. Off-Axis tensile test.
11.4 Experimental Determination of Engineering Constants
203
The measurement of the strain 45 in this direction allows to deduce the Youngs modulus in the 45 direction by:
E45 =
45 . 45
,
(11.60)
From Relation (11.9), this modulus is expressed by: 1 1 1 1 1 = + + 2 LT E45 4 EL ET GLT EL (11.61)
where only the in-plane shear modulus is unknown. This modulus is therefore deduced from the relation: 1 4 1 1 = + 2 LT . GLT E45 EL ET EL (11.62)
11.4.5 Practical Elements for the Tensile Tests

In practice, tensile tests are implemented using either dog bone specimens or straight sided specimens with end tabs (Figure 11.10). The load is usually applied to the ends of the test specimens by means of pin-type or serrated-jaw-type connections, in such a way to obtain a uniform distribution of the stresses in the gauge length of the specimens. The ends of pin-type connections tend to fail at low loads by shear. The dog bone specimens generally lead to initiation of cracks in the vicinity of the curvatures of the specimens because of stress concentrations, leading to errors in evaluating the fracture properties of materials. Thus, it is simplest to use serrated-jaw-type connections (pneumatic jaws or sel-clamping jaws) with straight sided specimens. The applied load is usually measured by means of a load cell provided with the testing device. The strains can be measured either by means of an extensometer or an electrical resistance strain gauge. Extensometers, mechanically attached to the test specimens, are quite simple to use. It is, nevertheless, necessary to watch for the absence of slipping when they are used. The strain gauges are more delicate to process and may be used for more accurate measurements of strains. In a longitudinal tensile test, it is necessary to ensure that the direction of the applied load does coincide with the fibre direction or the warp direction. In the case of unidirectional material or unbalanced cloth composite in the warp direction, misalignment by only a few degrees may result in measured values that are clearly lower than the actual values of the longitudinal Youngs modulus and of the ultimate strength. This problem is not as critical for transverse tensile tests. The application of the load by means of serrated-jaw-type connections produces in the clamped ends a stress state which is not exactly a uniaxial tensile stress state. At the level of the specimen ends a longitudinal tensile strain xx is induced when the transverse and shear strains are zero (yy = 0, xy = 0). The actual
204
(a)
(b)
FIGURE 11.10. Tension test specimens: a) dog bone form; b) straight sided with end tabs.
strain state thus differs from the uniaxial tensile state (11.6). However, according to the Saint-Venants principle, the strain and stress state is effectively a uniaxial tensile state far away from the specimen ends. Furthermore, in the case of off-axis tensile tests, there exists a coupling between in-plane tension and shear, which results from the terms 16 of the stiffness matrix. This coupling induces an Sshaped deformation of the test specimens (Subsection 15.1.1.3 and Figure 15.2). Therefore, to overcome these difficulties, it is necessary to use test specimens that are long compared to their width and to restrict the measurements to the central part of the test specimens.
EXERCICES
11.1 Plot (first in Cartesian coordinates and then in polar coordinates) the engineering constants Ex, Gxy, xy,x and xy as functions of the orientation in the cases of: a unidirectional composite:
LT = 0.31,
GLT = 4.5 GPa ;
an orthotropic composite:
EL = ET = 25 GPa,
LT = 0.12,
GLT = 4 GPa .
Exercises
205
11.2 With the help of the results established in Exercises 5.3 and 6.2, find directly the relations between the reduced stiffness matrix expressed in the direction and the reduced stiffness matrix in the material directions of an orthotropic composite. 11.3 Calculate the reduced stiffness constants in the material directions of the composites considered in Exercise 11.1. Next, calculate the reduced stiffness constants in a direction at 30 to the material directions. 11.4 Implement a numerical procedure having:
as inputs: the moduli EL, ET, LT, GLT and the orientation of a layer ; as outputs: the reduced stiffness constants Qij in the material directions, in the direction . and the off-axis constants Qij Apply this procedure to establish the results obtained in Exercise 11.3.
11.5 Further to the preceding procedure, implement a numerical procedure having:
as inputs: the strain state (xx, yy, xy) at a point of a layer; as outputs: the stresses (xx, yy, xy) referred to the geometric directions and the stresses (L, T, LT) in the material directions of the layer. Apply this procedure to the case of the layers considered in Exercise 11.3 and for the strain state:
xx = 1.5 %,
yy = 1 %,
xy = 2 % .
CHAPTER 12
Fracture Processes and Damage of Composite Materials
12.1 FRACTURE MECHANISMS INDUCED IN COMPOSITE MATERIALS 12.1.1 Introduction
Fracture mechanisms include every mechanical process producing a local discontinuity inside the body of a material, called a crack. It is usual to speak of initiation and propagation of fracture. The initiation of fracture can be considered as the nucleation of microcracks at the microscopic level (the level of the constituents) starting from defects. This process is known as microcracking. The fracture propagation is the result of the creation of new fracture surfaces at the macroscopic level (several times that of the constituents) developing from the existing microcracks. We shall also speak of macrocracking. In the case of composite materials, the fracture initiation is generally induced well before the observation of a change in the macroscopic behaviour of the materials.
12.1.2 Fracture Mechanisms Induced in a Unidirectional Composite Material

The final rupture of a unidirectional composite is the result of the accumulation of different elementary mechanisms: fibre fracture, transverse fracture of matrix, longitudinal fracture of matrix, fracture of the fibre-matrix interface.
12.1 Fracture Mechanisms Induced in Composite Materials
207
f = fu
fibre fracture
FIGURE 12.1. Fibre fracture.
Generally a mechanism is not isolated, but different mechanisms can be induced simultaneously. These mechanisms develop according to the nature of the materials and the conditions of the mechanical loading imposed. In a unidirectional fibre composite subjected to tensile loading in the fibre direction, fibre fracture occurs when the tension stress f in a fibre reaches the ultimate stress fu of the fibre (Figure 12.1). The fibre fracture produces a stress concentration in the vicinity of the fracture. The redistribution of the stresses, and consequently the resulting fracture process, depends principally upon: the fracture stress of the fibres, the capacity of the matrix to absorb the energy released, the properties of the fibre-matrix interface, etc. Figures 12.2a-d show the different processes of matrix fracture associated with fibre fracture.
(a)
(b)
(c)
(d)
FIGURE 12.2. Different modes of matrix fracture associated with fibre fracture: (a) Transverse fracture of matrix; (b) Shear fracture of matrix; (c) Debonding of fibre-matrix interface; (d) Longitudinal fracture of matrix.
208
Chapter 12 Fracture Processes and Damage of Composite Materials
m = mu
transverse cracking
FIGURE 12.3. Transverse fracture of matrix.
The fracture of matrix can be produced either by transverse cracking (Figure 12.3) when the tension stress m in the matrix reaches the ultimate stress mu of the matrix, or by longitudinal cracking (Figure 12.4) when the shear stress m in the matrix reaches the ultimate shear stress mu, generally in the vicinity of a fibre. This latter mode of fracture, called splitting, is produced when the debonding stress is greater than the ultimate shear stress of the matrix: d > mu. In the opposite case in which d < mu, debonding fracture is induced at the fibre-matrix interface (Figure 12.5). The final rupture of a unidirectional fibre composite is the result of the accumulation of these different elementary mechanisms. The initiation and then the propagation of the fracture are depending upon the properties of the fibres and
m = mu
longitudinal fracture
FIGURE 12.4. Longitudinal fracture of matrix.
209
m = d
debonding
FIGURE 12.5. Fibre-matrix debonding.
of the matrix, of the fibre-matrix interface, on the volume fraction of the fibres and on the mechanical loading conditions imposed.
12.1.3
Unidirectional Composite Subjected to a Longitudinal Tension
In the case of a unidirectional fibre composite subjected to a longitudinal tension, the fracture initiation is generally induced either by fracture of fibres when the fracture strain of the fibres is less than that of the matrix (fu < mu), or by transverse fracture of the matrix in the opposite case. In the case fu < mu, the stress-strain curves are as shown in Figure 12.6. Assuming the equality of the strains in the fibres and the matrix, Relation (9.80) is written at the instant of the fracture as:
cu = fuVf + ( m )
fu
(1 Vf ) ,
(12.1)
where cu is the ultimate stress of the composite, fu the ultimate stress of the fibres and ( m ) fu the stress in the matrix for a strain equal to the ultimate strain fu of the fibres. The stress ( m ) fu is less than the stress mu at the matrix fracture. Hence:
cu fuVf + mu (1 Vf ) .
(12.2)
Usually, the expression used for the ultimate stress of the composite material is the law of mixtures: cu = fuVf + mu (1 Vf ) , (12.3) with, for usual fractions of the fibres :
cu fuVf .
(12.4)
210
fu
stress
fibre
cu = fuVf + ( m ) (1 Vf )
fu
composite
mu
matrix
( m ) fu
fu
strain
mu
FIGURE 12.6. Stress-strain curve of a unidirectional fibre composite subjected to a longitudinal tension, in the case fu < mu.
fu
stress
fibre
( f ) mu
cu = ( f ) mu Vf + mu (1 Vf )
mu
matrix composite
mu fu strain
FIGURE 12.7. Stress-strain curve of a unidirectional fibre composite subjected to a longitudinal tension, in the case fu > mu.
211
In the case where the strain at the matrix fracture is less than that of the fibres (Figure 12.7), the ultimate stress of the composite is given by the expression:
cu = ( f )
mu
Vf + mu (1 Vf ) ,
(12.5)
where ( f ) mu is the stress in the fibre at the instant of the matrix fracture. The value of the fracture stress of the composite material is then strictly less than that given by Expression (12.1). In this case, the matrix does not allow to take really advantage of the fibre reinforcement. The mechanical characteristics (Chapter 2) of the usual fibres are: for carbon fibres: with high strength (HS fibres): Ef = 220 GPa, fu = 3,000 to 4,000 MPa, thus fu = 1.4 to 1.8 % ; with high modulus (HM fibres) : Ef = 400 GPa, fu 2,200 MPa, thus fu = 0.5 % ; for E-glass fibres: Ef = 70 GPa, fu = 2,400 to 3,400 MPa, thus fu = 3.4 to 4.8 %. The ultimate strain of the usual matrices is: for rigid polyesters: for phenolic resins: for epoxide resins:
mu = 2 to 5 %, mu 2.5 %, mu = 2 to 5 %.
In addition, in industrial use, the resins are greatly filled, leading to a significant decrease of the ultimate strain of the matrix. The previous values therefore show that the fracture properties of high-performance composites (carbon fibre composites) are governed by the properties of the fibres. In contrast, in the case of industrial composites with glass fibres, the fracture properties may be limited by too low a strain of the matrix. It would thus appear to be necessary to have the best adaptation of the properties of the matrix to those of the fibres to optimize the fracture properties of the composite materials. After initiation, the fracture propagation differs according to the nature of the fibre-matrix interface. In the case of high fibre-matrix bonding, the fracture, initiated either by fibre fracture or by matrix fracture, induces a high stress concentration near the crack tip that leads to successive propagation of the fracture in the fibres and in the matrix (Figure 12.8a). The fracture observed is of brittle type (Figure 12.8b). It is also possible to observe a bridging, by longitudinal fracture of the matrix or by fibre-matrix debonding, of two cracks developing in different zones (Figures 12.8c and 12.8d). In the case of poor fibre-matrix bonding, the transverse propagation of cracking develops in the following way. Near the crack tip, the high stress concentration induces shear fracture of the fibre-matrix interface, and the crack propagates transversely to the fibres (Figure 12.9), without fibre fracture. Behind the crack tip, the crack opening induces high tension stresses in the fibres bridging the crack planes. These high tension stresses lead to the fracture of fibres at a distance more or less next to the crack planes, according to the defects in the fibres.
212
(a)
(b)
(c)
(d)
FIGURE12.8. Fracture propagation in the case of high fibre-matrix bonding.
fibre-matrix debonding
direction of fracture propagation
fibre pulling-out
debonding
FIGURE12.9. Fracture propagation in the case of poor fibre-matrix bonding.
213
FIGURE 12.10. Fracture surfaces in the case of poor fibre-matrix bonding.
As the crack continues to propagate, the broken fibres are pulled out of the matrix. In some cases, on reaching the fibre-matrix interface, the crack can split and propagate along the fibres. Thus, different types of fracture surfaces can be observed (Figure 12.10), with a brush aspect of the crack surfaces.
12.1.4 Fracture of a Unidirectional Composite Under Transverse Tensile Loading

In the case of a unidirectional fibre composite subjected to transverse tensile loading, the fracture occurs either by matrix fracture or by debonding of the fibrematrix interface. Matrix fracture is induced when the tensile stress m in the matrix reaches the ultimate stress mu of the matrix (Figure 12.11). This process occurs when the fracture stress of the matrix is lower than the tensile debonding stress d of the fibre-matrix interface. In the opposite case where mu > d, the fracture
214
m = mu < d
FIGURE 12.11. Fracture of a unidirectional fibre composite subjected to transverse tensile loading.
of the unidirectional fibre composite is produced by fracture of the fibre-matrix interface. After initiation, fracture propagates inside the composite material through a fracture surface which is more or less plane, according to the properties of the composite material.
12.1.5 Laminate Fracture Modes

In the case of laminates, besides the basic mechanisms described previously (fibre-matrix debonding, longitudinal fracture of the matrix, transverse fracture of the matrix, fibre fracture), another fracture mode may be observed called delamination (Figure 12.12), which consists of the separation of layers from one another. The fracture processes induced depend upon the nature of the constituents, the architecture of the laminates and the type of mechanical loading imposed to the laminate. For example, in the case of a cross-ply laminate subjected to tensile loading in the 0 direction (Figure 12.13), the first process of fracture which is observed is the fracture of the layers with 90 orientation. Fracture is induced by longitudinal cracking of the matrix or by fracture of the fibre-matrix interface in the 90 layers. This fracture process leads to the development of cracks (Figure 12.14) which are transverse to the direction of the mechanical loading. Thus, this initial fracture process in the 90 layers is called the transverse cracking of the cross-ply laminate. When the mechanical loading is increased, the crack numbers increases up to a saturation state of the cracking. The transverse cracks induce at the crack tips, between the 90 and 0 layers, stress concentrations which lead to the initiation and then to the propagation of delamination at the interface between the 0 and 90 layers. This delamination process develops next up to the final rupture of the cross-ply laminate which occurs by fracture of the fibres and the matrix in the 0 layers. Figure 12.14 shows the final aspect of the fracture surfaces after rupture of the laminate.
215
fibre-matrix debonding
longitudinal fracture transverse fracture of of matrix matrix fibre fracture
delamination
FIGURE 12.12. Fracture mechanisms observed in laminates.
90
90
Figure 12.13. Cross-ply laminate subjected to tensile loading in the 0 direction.
In the case of a 45 angle-ply laminate subjected to longitudinal tension in the 0 direction (Figure 12.15), there is first observed the longitudinal fracture of the matrix is the 45 layers, followed by the delamination interface between the layers. Figure 12.16 gives an example of the fracture surfaces observed in this case. Another interesting example is that of a plate constituted of a [0, 45, 90]n laminate with a hole at its centre and subjected to tensile loading in the 0 direction (Figure 12.17a). Several stages of cracking are observed in this case. In the first stage, the longitudinal cracking of the matrix is induced in the 90 layers (Figure 12.17b). In the second stage, matrix cracking is initiated in the 45 layers, from the cracks propagated in the 90 layers, with a limited propagation of the cracks at 45 (Figure 12.17c). The third stage is characterized by the initiation of longitudinal matrix cracks initiated from the hole, which propagate in
216
FIGURE 12.14. Fracture of a [0/90]2S fibre carbon composite with a hole at its centre. (top) Macroscopic fracture near the hole; (bottom) Edge of the test specimen far from the hole: transverse cracking in the 90 layers (ONERA document).
217
45
45
45
45
FIGURE 12.15. 45 angle-ply laminate subjected to tensile loading in the 0 direction.
the 0 layers. These cracks also generate secondary cracks in 45 layers (Figure 12.17d). In the last stage, the longitudinal cracks in the 0 layers induce a delamination of the layers, followed by the fracture of the 90 layers, then of the 45 layers, and lastly by the fracture of the fibres in the 0 layers, leading to the final fracture of the plate.
FIGURE 12.16. Fracture of a [45]2S carbon fibre composite (ONERA document).
218
(a)
(b)
(c)
(d)
FIGURE 12.17. Progressive cracking of a [0, 45, 90]n laminate. (a) Plate with a hole at its centre subjected to tensile loading; (b) 1st stage: cracking in 90 layers; (c) 2nd stage: cracking in 45 layers; (d) 3rd stage: fracture in 0 layers.
12.1.6 Observation of Fracture Mechanisms

The observation of the fracture mechanisms in laminates can be carried out by different techniques. We give hereafter some basic elements on these techniques.
12.1.6.1 Observation by Microscopy

Optical observation with a microscope is a very simple technique to carry out for the continuous observation of fracture mechanisms during tests. However, this technique is restricted to local observation and the depth field is limited. Scanning electronic microscopy increases the depth field, allowing high magnifications to be obtained. Figures 12.18 and 12.19 show the micrographs obtained in the case of transverse fracture of composites with poor fibre-matrix bonding (Figure 12.18) and high bonding (Figure 12.19).
12.1.6.2 Radiography Analysis

The technique of analysis by X-radiography consists in impregnating the test specimens by means of opacifying agent (as zinc iodide) and then taking an Xradiograph of the test specimens. Radiography gives a two-dimensional image of the fracture state (Figure 12.20). It is, however, easy to localize the damage in
219
FIGURE 12.18. Fracture surface associated with poor fibre-matrix bonding in the case of a carbon fibre composite (ONERA document).
220
FIGURE 12.19. Fracture surface associated with high fibre-matrix bonding in the case of a carbon fibre composite (ONERA document).
221
plain specimens
specimens with a hole
impacted specimens
[0/90]2S
[0/45/90]2S
[90/45/0]2S
FIGURE 12.20. X-ray observation of the fracture state of carbon fibre composites after fatigue (105 cycles; R = 0.1) in the case of different test specimens: plain specimens; specimens with a hole (diameter of 5.6 mm) at its centre; impacted specimens (projectiles of 5.56 mm diameter at a speed of 1,000 m/s) and with different stacking sequences of the laminates: [0/90]2S , [0/45/90]2S , [90/ 45/0]2S (ONERA document).
222
the body of laminates when one knows the orientations of the layers. Radiography allows a very fine observation of the cracks, and of the cracks transverse to the thickness of laminates in particular. It should be noted that it is necessary to demount the test specimen for each radiography, and then to remount it in the testing machine in order to carry on the test. This makes the tests considerably time consuming. It is also possible to observe the fracture state of test specimens by radiography with a medical scanner. The analysis of the density variations allows us to obtain information in three dimensions.
12.1.6.3 Acoustic Emission Analysis

The preceding techniques permit observations at different times. They are also time consuming to carry on because of the mounting and demounting of the test specimens, necessary for the observations on the fracture state. In contrast, acoustic emission is a physical process which allows us to access, in real time, information about the fracture mechanisms as they happen. When a fracture mechanism is induced inside a material, it creates a local discontinuity of the displacement and stress fields. This discontinuity, called an event, generates a strain wave which propagates through the material. At the surface of the material, an adapted transducer converts the wave received (Figure 12.21) into an electric signal (the acoustic emission signal) which is next amplified, then analysed. The transducers are piezoelectric transducers, developed specifically for acoustic emission so that they have a high sensitivity. The frequency domain studied generally extends from 50 kHz to 1 MHz. Figure 12.22 gives examples of acoustic emission signals. The technique of acoustic emission consists in extracting from the signals information about the fracture mechanisms. The analyses used are:
amplification acoustic emission transducer acoustic emission signal
material
fracture process (event)
FIGURE 12.21. The acoustic emission process.
12.2 Failure Criteria
223
1 0
Normalized amplitude ( 1 = 50 mV )
-1 0
1 0
50
100
150
200
250
300
350
400
450
500
-1 0
1 0 -1 0 1 0 -1 0
50
100
150
200
250
300
350
400
450
500
50
100
150
200
250
300
350
400
450
500
50
100
150
200
250
300
350
400
450
500
Time ( s ) FIGURE 12.22. Acoustic emission signals recorded during bending tests on unidirectional carbon fibre-epoxide composites.
counting the signals, which shows when the fracture processes are initiated, and allows us to obtain information about cracking activity; localization of the damage, by measuring the times of arrival of the signals at several transducers suitably situated on the test specimens; frequency analysis of the signals; amplitude analysis, which consists of measuring the peak amplitude of each signal, then analyzing the evolution of the statistical distribution of the amplitudes during tests. etc. The processes of recording and analysis of the acoustic emission signals are greatly improved by the numerical equipment which is available today for the engineer.
12.2 FAILURE CRITERIA 12.2.1 Introduction

The objective of the failure criteria is to allow the designer to have an evaluation of the mechanical strength of laminates. Quite generally, the mechanical resistance of a material is associated to an irreversible degradation: for example the actual fracture of the material (Figure 12.23a) or the end of the
224
elastic domain (Figure 12.23b). In fact, the definition of failure may change from one application to another. In the case of composite materials, the end of the elastic domain is generally associated with the development of microcracking: matrix microcracking, fibre-matrix debonding, etc. In the initial stage of fracture process, the initiated microcracks do not propagate, and their development changes the stiffness of the material very gradually. Failure criteria have been established in the case of a single layer of a laminate and may be classified as: the criterion of maximum stresses, the criterion of maximum strains, the interactive criteria, usually called as energy criteria.
12.2.2 Maximum Stress Criterion

12.2.2.1 Criterion in the Material Directions
The maximum stress criterion introduces: Xt, Xc : the tensile and compressive strengths in the longitudinal direction, respectively, Yt, Yc : the tensile and compressive strengths in the transverse direction, respectively, S : the in-plane shear strength of the layer. The longitudinal and transverse directions are the material directions of the layer under consideration (Figure 12.24). The strength quantities are the positive values of the fracture stresses measured in tensile, compressive and shear tests. In the case of a layer subjected to a plane stress state (L, T, LT) expressed in the material directions, the maximum stress criterion considers that the fracture of the layer occurs when one of the stresses (L, T, LT) has reached the corresponding value of the strength.
fracture
elasticity limit
(a) (b) FIGURE 12.23. Brittle (a) and ductile (b) behaviour of a material.
225
LT T
L
L
T
LT LT LT T
L
FIGURE 12.24. Stresses in the material directions of a layer.
Thus, the maximum stress criterion can be written in the form:

Xc < L < Xt , Yc < T < Yt , S < LT < S .
(12.6)
If these six inequalities are satisfied, then the failure of the layer does not occur. If anyone of these inequalities becomes not satisfied, then the layer failure occurs by the fracture mechanism corresponding to the stress (Xt, Xc, Yt, Yc or S) of the inequality that is not satisfied
12.2.2.2 Magnitude Orders of the Fracture Stresses

The values of the fracture stresses are measured in tensile, compressive and shear tests. In practice, the experimental investigation may lead to some difficulties associated with the anisotropy and heterogeneity of materials. In particular, premature fracture of the test specimens can occur in failure modes which are not wanted. For example, compressive test specimens may fail by shear process or buckling. Moreover, pure shear tests are difficult to implement. Other problems can be related to the methods of fabrication of the test specimens. Finally, the available experimental values are limited and the problem of the evaluation of the fracture stresses stays opened. We give some elements in the case of unidirectional fibre composites. In the case where the ultimate strain of the matrix is higher than that of the (Subsection 12.1.3), the longitudinal tensile strength of a unidirectional composite is given by the law of mixtures (12.1), that is:
X t = fuVf + ( m )
fu
(1 Vf ) ,
(12.7) (12.8)
with, for the usual proportions of the fibres: X t fuVf .
226
Xt
1400 1200 1000 800 600 400 200 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
fu = 2500 MPa
fu = 1500 MPa
Vf
FIGURE 12.25. Longitudinal tensile strengths of unidirectional glass fibre composites as function of fibre volume fraction.
In practice, it is quite difficult to have accurate values of fu. For example, at the exit of the bushing, E-glass fibres have a fracture stress of the order of 3,500 MPa. This value decreases as a result of the handling and chemical attacks to which the fibres are submitted up to their incorporation in the matrix. At the stage of the moulding process, the values of the fracture stresses of fibres are estimated to be of the order of 1,500 to 2,500 MPa. Figure 12.25 gives an evaluation of the longitudinal strengths of unidirectional glass fibre composites for fibre volume fractions lying between 0.2 and 0.7. The values measured for the fracture stresses in longitudinal compression depend on the nature of the fibres and the fibre-matrix interface. In the case of a transverse tensile test on a unidirectional composite, the fracture stress (Subsection 12.1.4) corresponds to the weakest link: the matrix or the fibre-matrix interface. The values of the transverse tensile strength Yt are usually less than the fracture stress of the matrix, and varies little with the proportion of the fibres. Taking into account the dispersion observed in fracture tests, it is usual to consider that this value is constant with the fibre fraction, with values lying between 20 and 60 MPa. In contrast, the transverse fracture stress Yc in compression is higher, at about 100 to 150 MPa. The shear fracture stress S is a parameter that is quite difficult to evaluate. The experimental results show that S does not depend in practice upon the proportion of the fibres and is of the same order of magnitude as the shear fracture stress of the matrix. According to the type of matrix and the quality of the fibre-matrix interface, the shear fracture stress is about 40 to 80 MPa. Table 12.1 gives examples of values of fracture stresses measured on epoxide matrix composites: three unidirectional composites and a balanced cloth composite. The values reported in this table have to be considered as indicative.
227
TABLE 12.1. Typical fracture characteristics measured for various composites with epoxide matrix.
Unidirectional composites Fibres

Vf Xt Xc Yt Yc S
Balanced cloth Carbon 0.4 500 350 460 350 50
E-glass 0.60
HS Carbon 0.60 1,380 1,430 40 240 70
Kevlar 0.60 1,400 280 15 50 35
(MPa) (MPa) (MPa) (MPa) (MPa)
1,400 910 35 110 70
12.2.2.3 Off-Axis Failure Criterion

In applications, the stresses are referred to the reference system (x, y, z) of the laminate (Figure 12.26). To apply the failure criterion, the stresses xx, yy and xy in a layer must be transformed to stresses in the material directions of the layer from Relation (5.44). Whence:
L = xx cos 2 + yy sin 2 + 2 xy sin cos , T = xx sin 2 + yy cos 2 2 xy sin cos , LT = ( yy xx ) sin cos + xy ( cos 2 sin 2 ) ,
z T T
(12.9)
xy yy xy
xx
y
yy xy
xx
xy
L
FIGURE 12.26. Layer referred to the reference system of the laminate.
228
and the fracture criterion (12.6) may be expressed as:

X c < xx cos 2 + yy sin 2 + 2 xy sin cos < X t , Yc < xx sin 2 + yy cos 2 2 xy sin cos < Yt , S < ( yy xx ) sin cos + xy ( cos 2 sin 2 ) < S . (12.10)
12.2.2.4 Off-Axis Tension or Compression

In the case of off-axis tension or compression (Figure 12.27), the stresses (12.9) reduce to:
L = xx cos 2 , T = xx sin 2 , LT = xx sin cos ,

and the maximum stress criterion is expressed as: (12.11)
X c < xx cos 2 < X t , Yc < xx sin 2 < Yt , S < xx sin cos < S . This criterion can be represented graphically by plotting the maximum value xu of the tensile or compressive stress xx, for which one of the criteria is reached, as a function of the angle between the loading direction and the longitudinal material direction of the material. In a tensile test, the tensile stress xu corresponds to the smallest of the values: (12.12)
xx
xx
FIGURE 12.27. Off-axis tension.
229
xu =
Xt cos 2
xu =
Yt sin 2
xu =
S , sin cos
(12.13)
and in a compressive test, the compressive stress xu corresponds to the smallest of the values: Xc Y S (12.14) xu = , xu = c2 , xu = , 2 sin cos cos sin the value xu being then the positive determination of the stress. Figure 12.28 shows the results obtained in the case of a unidirectional E-glass fibre composite the fracture characteristics of which are given in Table 12.1. The scale adopted for the values of the stress xu is logarithmic, to expand the scale for the low values. We observe a steep decrease of xu with the angle . In a tensile test, the value of xu = 1,400 MPa for angles close to 0 is no more than the order of 200 MPa for an angle of 25.
2000
1000
tension X t / cos 2
Fracture stress xu ( MPa )
700 400
compression X c / cos 2
S / sin cos
compression X c / sin 2 tension X t / sin 2
200
100 70 40
20
10
20
30
40
50
60
70
80
90
Fibre orientation ( )
FIGURE 12.28. Maximum stress criterion in the case of a unidirectional glass fibre composite.
230
12.2.3 Maximum Strain Criterion

12.2.3.1 Criterion in Material Directions
The maximum strain criterion is quite similar to the maximum stress criterion, because the strains are limited instead of the stresses. The maximum strain criterion introduces: X t (X c ): the ultimate tensile (or compressive) strain in the longitudinal direction, Y t (Y c ): the ultimate tensile (or compressive) strain in the transverse direction, S : the ultimate in-plane shear strain of the layer. The layer is considered to have failed if one of the strains ( L , T , LT ) in the material directions have reached the corresponding ultimate strain. The maximum strain criterion is then written in the form:
Xc < L < X t , Y c < T < Y t , S < LT < S .
(12.15)
12.2.3.2 Off-Axis Tension or Compression

In the case of off-axis tension or compression (Figure 12.27), the stresses in the material directions are given by Relation (12.11). In the case of a plane stress state, the strains in the material directions are expressed as:
L S11 = S T 12 LT 0 S12 S 22 0 0 L 0 T . S66 LT
(12.16)
So, by associating Relations (12.11) and (12.16) we obtain:
( ) T = ( S12 sin 2 + S22 cos 2 ) xx ,

L = S11 cos 2 + S12 sin 2 xx , LT = S66 sin cos xx .
(12.17)
The compliance stiffnesses Sij are expressed as functions of the engineering moduli determined in the material directions EL, ET, LT, TL and GLT, according to Relations (9.31) in the case of unidirectional composites and according to Relations (10.9), (10.13) and (10.20) in the case of orthotropic composites. Relations (12.17) which express the strains are then transformed as:
231
1 ( cos2 LT sin 2 ) xx , EL 1 ( sin 2 TL cos2 ) xx , T = ET 1 sin cos xx . LT = GLT
L =
(12.18)
The maximum strain criterion must lead to values identical to those found with the maximum stress criterion in the case of longitudinal tension (or compression): = 0, and in the case of transverse tension (or compression): = 90. This implies that: X X X t = t , X c = c , EL EL (12.19) Yt Yc Y t = Y c = , . ET ET
Furthermore, the identity of the shear fracture criterion in both cases leads to: S = S . GLT
Xt cos 2 LT sin 2 Yt sin LT cos 2
2
(12.20)
From this it results that the maximum strain criterion (12.15) can be rewritten as:
Xc cos 2 LT sin 2 Yc sin LT cos
2 2
< xx < < xx <
, ,
(12.21)
S S . < xx < sin cos sin cos
By comparing these expressions with Expressions (12.12) obtained in the case of the maximum stress criterion, we observe that the two criteria differ simply in the introduction into the maximum strain criterion of terms that are functions of the Poisson ratios LT and TL. These terms in practice modify the numerical results slightly. The ultimate stresses are modified in the same way. In a tensile test, the ultimate stress xu corresponds to the smallest of the values:
xu = xu = xu =
Xt , cos LT sin 2
2
Yt sin LT cos 2 S , sin cos

2
(12.22)
232
and in a compressive state, the positive determination of the ultimate stress xu corresponds to the smallest of the values:
xu = xu = xu =
Xc cos LT sin 2 Yc sin TL cos 2 S . sin cos

2 2
, , (12.23)
As in the case of the maximum stress criterion, it is possible to plot the ultimate stress xu as a function of the angle . The curves obtained differ slightly from those obtained with the maximum stress criterion (Figure 12.28).
12.2.3.3 Comparison between Maximum Stress and Strain Criteria

The results of the previous subsection show a similarity between the two criteria of maximum stress and maximum strain. To go more thoroughly into the comparison between these two criteria, we consider the example of a layer loaded in a plane stress state (Figure 12.29) such as:
L = 12 T
and
LT = 0 .
(12.24)
The layer is constituted of a unidirectional E-glass fibre composite the fracture characteristics of which are given in Table 12.1 and the engineering constants of which are reported in Table 9.2. So:
X t = 1,400 MPa, Yt = 35 MPa, S = 70 MPa, EL = 46 GPa, ET = 10 GPa, GLT = 4.6 GPa, LT = 0.31. We look for the values of stresses L and T (L = 12T) for which fracture occurs. T
T L L = 12 T
T
FIGURE 12.29. Layer loaded in a particular state of plane stresses.
233
1. Application of Maximum Stress Criterion The criterion (12.6) of maximum stresses is written here as:
L < Xt,
Thus: 12 T < X t T < Yt , or
T < Yt .
1 T < 12 X t = 117 MPa T < Yt = 35 MPa.
The value of the ultimate stress is given by the smallest of the two values. It results that fracture occurs by transverse fracture. The stress state is then:
L = 12 35 = 420 MPa, T = 35 MPa.

2. Application of Maximum Strain Criterion
(12.25)
According to (12.16), the strains in the material directions may be written as:
1 ( L LT T ) , EL 1 T = S12 L + S22 T = ( TL L + T ) . ET
L = S11 L + S12 T =
(12.26)
Assuming that the behaviour of the material is linear up to fracture, the fracture strains are expressed by (12.19) and the maximum strain criterion (12.15) may thus here be written as:
L LT T < X t , TL L + T < Yt .
Therefore, since L = 12 T :
(12.27)
L < T
Xt = 120 MPa, 12 LT Yt < = 183 MPa, 1 12 TL
(12.28)
considering Relation (9.27) to derive the Poisson ratio TL as function of the other moduli. The value of the ultimate stress is given by the lowest of the two values. It follows that fracture occurs by longitudinal fracture. The stress state is then:
L = 1,440 MPa,
T = 120 MPa .
(12.29)
The values obtained in (12.25) and (12.29) show the contradictory results in which the two apparently similar theories result: the values differ by a factor of 3.43 and the fracture mode is changed: longitudinal fracture in one case and transverse fracture in the other. This contradiction in fact lies in the misuse done to establish the relation between the value of the ultimate stresses and the ultimate strains. In practice,
234
these values would have to be measured respectively in the case of plane stresses and of plane strains. The respective criteria which would then be deduced from them could be applied to these states. In this way, the relations between the ultimate stresses and strains would be more complex.
12.2.4 Interactive Criteria

In some applications, the maximum stress and strain criteria do not allow us to describe the experimental results observed. Moreover, these criteria do not consider interactions between the modes of fracture: longitudinal, transverse and shear fracture. So, the different fracture mechanisms are assumed to occur independently. Thus, interactive criteria have been investigated, extending to the case of orthotropic materials the Von Mises criterion introduced for the isotropic materials. Von Mises criterion is related to the deformation energy stored per unit volume of the strained material. This is the reason why these interactive criteria are sometimes called energy criteria. However, in the case of orthotropic materials, these criteria are not related to the strain energy exclusively.
12.2.4.2 Hills Criterion

One of the first interactive criteria for fracture applied to anisotropic materials was introduced by Hill [15]. This criterion can be formulated by stating that the fracture of an anisotropic material does not occur as long as the following inequality is satisfied:
F ( T T ) + G ( T L ) + H ( L T )
2 2 2 + 2 L TT 2 + 2 M LT 2 + 2 N LT 2
< 1.
(12.30)
The fracture of the material thus happens when the equality is satisfied. That is:
F ( T T ) + G ( T L ) + H ( L T )
2 2 2 + 2 L TT 2 + 2 M LT 2 + 2 N LT 2
= 1.
(12.31)
This equality is Hills criterion referred to the material directions (L, T, T'). It can also be put into another form as follows:
2 2 2 + ( F + H )T + ( F + G ) T (G + H ) L 2 H L T 2G L T 2 2 2 2 F T T + 2 L TT + 2 M LT + 2 N LT = 1.
(12.32)
The quantities F, G, H, L, M and N are characteristic parameters of the material under consideration, and which are related to the fracture stresses X, Y and S of the material according relations that we establish hereafter.
235
In the case of a tensile (or compressive) test in the longitudinal direction L, Hills criterion is reduced to: 1 (12.33) G+H = 2 , X where X is the tensile (or compressive) strength in the direction L. Similarly we find: 1 (12.34) F+H = 2 , Y
F +G = 1 Z2
(12.35)
where Y and Z are the tensile (or compressive) strengths in the directions T and T'. In the case of a shear test in the plane (L, T), Hills criterion is reduced to:
2N = 1
2 S LT
(12.36)
where SLT is the shear strength in the plane (L, T). Similarly we have:
2M = 1
2 S LT
(12.37) (12.38)
2L =
1
2 STT
where SLT' and STT' are the shear strengths in the respective planes (L, T') and (T, T'). Expressions (12.33) to (12.38) allow us to derive the fracture parameters F, G, L, M, N and to write Hills criterion in the form: 1 1 L T T 1 + + 2 + 2 2 L T X Y Z X Y Z 1 1 1 1 1 1 2 + 2 2 L T 2 + 2 2 T T X Y Z Y Z X + LT + LT + TT = 1. S LT S LT STT It is to be noted that Hills criterion does not take into account the difference between the behaviour of the materials under tension and compression. In the case of a plane stress state in the plane (L, T) of the material layer, we have: T' = LT' = TT' = 0, and Hills criterion simplifies as: 1 1 LT L T 1 + 2 + 2 2 L T + =1. X Y X Y Z S LT
2 2 2 2 2 2 2 2 2
(12.39)
(12.40)
236
12.2.4.3 The Tsai-Hill Criterion

The preceding fracture criterion (12.40) under plane stresses has been simplified by Azzi and Tsai [16] in the case of unidirectional composite materials. In fact, in this case we have: Z = Y, and the criterion (12.40) may be written as:
L T L T LT + + =1. X2 X Y S LT
(12.41)
This criterion is usually known as the Tsai-Hill criterion. In the case of tension or compression off the material directions (Figure 12.27), the stresses in the material directions are given by Expressions (12.11). Substituting these expressions in Relation (12.41), the Tsai-Hill criterion is written as:
cos 4 X2 1 1 sin 4 1 + 2 2 sin 2 cos 2 + = 2 . 2 Y xx S LT X
(12.42)
Figure 12.30 shows the results derived from the Tsai-Hill criterion in the case of unidirectional E-glass fibre composites with the same characteristics as in the case of Figure 12.28. There is generally a good agreement between these results and the experimental results obtained for this type of composite.
2000
tension
1000
Fracture stress xu ( MPa )
700 400
200
compression
100 70 40
20
10
20
30
40
50
60
70
80
90
Fibre orientation ( )
FIGURE 12.30. The Tsai-Hill criterion in the case of unidirectional glass fibre composite.
237
12.2.4.4 Hoffmans criterion

A generalization of Hills criterion that takes account of the difference between the behaviour of the materials under tension and compression was formulated by Hoffman [17]. Hoffmans criterion states that the material fracture occurs when the following equality is satisfied: C1 ( T T ) + C2 ( T L ) + C3 ( L T )
2 2 2
2 2 2 + C4 L + C5 T + C6 T + C7 TT + C8 LT + C9 LT = 1.
(12.43)
The constants C1 to C9 are characteristics of the material, and are related to the failure strengths of the material by the relations: 1 1 1 1 , C1 = + 2 YtYc Z t Z c X t X c 1 1 1 1 + , C2 = 2 Z t Z c X t X c YtYc 1 1 1 1 + , C3 = 2 X t X c YtYc Z t Z c 1 1 1 , C4 = C7 = 2 , Xt Xc STT C5 = C6 = 1 1 , Yt Yc 1 1 , Z t Zc C8 = C9 = 1
2 S LT 1 2 S LT
(12.44)
, .
In the case of a plane stress state in the plane (L, T), Hoffmans criterion is reduced to:
2 L
Xt Xc
2 T
YtYc
L T
Xt Xc
Xc Xt Y Y 2 L + c t T + LT =1. 2 Xc Xt YcYt S LT
(12.45)
12.2.4.5 Tsai-Wu General Theory

12.2.4.5.1 Formulation
The preceding criteria are usually sufficient for describing the various experimental results that are observed. However, one of the ways of improving the correlation between experimental and theoretical results is to increase the number of parameters in the prediction equations. This, associated with the ability of putting the failure criteria into tensor form, led Tsai and Wu [18] to postulate that fracture of an anisotropic material occurs when the following equality is satisfied:
238
Fi i + Fij i j = 1,
i, j = 1, 2, . . . , 6 ,
(12.46)
where the constants Fi and Fij are the components of two tensors of respective order 2 and 4. The usual contraction notation is used in this relation for stresses referred to the material directions:
1 = 11 = L ,
2 = 22 = T ,
3 = 33 = T ,
4 = 23 = TT , 5 = 13 = LT , 6 = 12 = LT .
Equation (12.46) may be written in the expanded form as:
F11 + F2 2 + F3 3 + F4 4 + F5 5 + F6 6
2 + F111 + 2 F121 2 + 2 F131 3 + 2 F141 4 + 2 F151 5 + 2 F161 6 2 + F22 2 + 2 F23 2 3 + 2 F24 2 4 + 2 F25 2 5 + 2 F26 2 6 2 + F33 3 + 2 F34 3 4 + 2 F35 3 5 + 2 F36 3 6 2 + F44 4 + 2 F45 4 5 + 2 F46 4 6 2 + F55 5 + 2 F56 5 6 2 + F66 6 = 1.
(12.47)
The linear terms take account of the possible difference between the behaviour of the material under tension and compression. The quadratic terms Fij define an ellipsoid in the stress space and take account of the interactions between the stresses i and j. The interest of the formulation developed by Tsai-Wu lies in: 1. the invariance of the form of Relation (12.46) under any change of basis; 2. the transformation of the criterion according to the transformation laws of the tensors i, ij or Fi, Fij ; 3. the symmetry properties of the tensors Fi, Fij similar to those of the elasticity constants.
12.2.4.5.2 Expression for the Constants
Here, we are interested in the case of an orthotropic composite material subjected to plane stresses in the plane (1, 2) = (L, T). Relation (12.47) is then be written as:
2 2 2 F1 1 + F2 2 + F6 6 + F11 1 + F22 2 + F66 6 + 2 F12 1 2 = 1 ,
or
2 2 2 F1 L + F2 T + F6 LT + F11 L + F22 T + F66 LT + 2 F12 L T = 1 .
(12.48)
The parameters Fi and Fij can be expressed in terms of the material strengths measured in different tests.
239
In the case of a tensile test in the direction L, the strength Xt is such that:
F1 X t + F11 X t2 = 1 ,
(12.49) (12.50)
and in a compressive test:

2 F1 X c + F11 X c =1.
From these two relations we deduce:

1 1 , Xt Xc 1 F11 = . Xt Xc F1 =
(12.51)
By analogy we have, similarly:

1 1 , Yt Yc 1 F22 = . YtYc F2 =
+ S LT
(12.52)
In the case of a shear test in the plane (L, T) (Figure 12.31a), the fracture stress is such that:
+ +2 F6 S LT + F66 S LT = 1.
(12.53)
On reversing the direction of the stresses (Figure 12.31b), the fracture shear stress S LT is such that:
2 F6 S LT + F66 S LT = 1.
(12.54)
These two relations lead to:

F6 = 1
+ S LT
1
S LT
, .
(12.55) (12.56)
F66 =
+ S LT S LT
The fracture stress is independent of the sign of the shear stress. So:
+ S LT = S LT = S LT .
(12.57)
It results that in the case of orthotropic materials we have:

F6 = 0 , F66 = 1
2 S LT
(12.58) . (12.59)
240
LT
LT
L
LT
LT
L
LT LT
LT
LT
(a)
(b)
FIGURE 12.31. Shear test.
It remains to determine the coupling parameter F12. This parameter can be determined in a biaxial experiment, for example biaxial tension. Such a test is implemented by applying the same value of the tensile stress in the directions 1 and 3 of the material. The stresses are then: 1 = 2 = , the other stresses being zero. Thus, the criterion (12.48) is written as:
( F1 + F2 ) + ( F11 + F22 + 2 F12 ) 2 = 1 .

Whence the expression for the interaction parameter:
F12 = 1 1 1 1 1 1 2 1 + + + 1 . 2 2 X t X c Yt Yc X t X c YtYc
(12.60)
(12.61)
The value of F12 corresponds to the value of the stress just measured at the instant of the fracture under biaxial tension. In practice, the interaction coefficient F12 can also be determined in a tensile (or compressive) test at 45 to the orthotropic material. In this case, the stresses in the material directions are:
1 = 2 = 6 =
45
2
3 = 4 = 5 = 0,
(12.62)
where 45 is the tensile stress applied. The criterion (12.48) may be written in this case:
2 45
( F11 + F22 + 2 F12 + F66 ) +
45
2
( F1 + F2 ) = 1 .
(12.63)
241
Whence the expression for the parameter F12 obtained in this test:
F12 =
2 45 1 1 1 1 1 + 2 2 X t X c Yt Yc 45 1 1 1 + + 2 . 4 X t X c YtYc S LT
2 45
(12.64)
The value of F12 corresponds to the value of 45 measured at the instant of the fracture in a tensile test at 45.
12.2.4.5.3 The Tsai-Wu Criterion for a Plane Stress State
Considering the preceding results, the Tsai-Wu criterion (12.48), for a plane stress state, is written in the form:
2 2 2 1 1 1 1 L T LT L T + + + + + 2 F12 = 1 , (12.65) L T X 2 X t X c YtYc S LT Xt Xc t Xc Yt Yc , expressed as: on introducing the coupling coefficient F12
= F12
1 X X X X 1 X c X t + t c (Yc Yt ) + 1 + t c 2 . (12.66) 2 YtYc YtYc 2

= F12
or
2 X X 1 X c X t + t c (Yc Yt ) 45 2 YtYc 45 2 X X X X 2 + 1 + t c + t2 c 45 , YtYc S LT 4
(12.67)
where and 45 are the fracture stresses measured in a biaxial test or in a tension
is considered as an at 45, respectively. Very often, the coupling coefficient F12 empirical coefficient, adjusted so as to fit the results deduced from (12.65) with the experimental results.
: In the case where the coupling coefficient is taken equal to 1 2

F12 =1 , 2
(12.68)
the Tsai-Wu criterion (12.65) for plane stresses is written as:

2 2 2 1 1 1 1 L T LT + + 2 L T = 1 . (12.69) L + T + X t X c YtYc S LT X t X c Xt Xc Yt Yc
We recover Hoffmans criterion (12.45) for plane stresses.
242
If, moreover, the tensile and compressive strengths are identical:

Xt = Xc = X , the criterion (12.69) becomes:
2 2
Yt = Yc = Y ,
2
(12.70)
L T LT L T =1. + + X Y S LT X2
This criterion is thus identical to the Tsai-Hill criterion (12.41).
(12.71)
EXERCISES
12.1 We consider an orthotropic layer the fracture characteristics of which are given by :
X t = 1,500 MPa, Yt = 90 MPa,
X c = 1,700 MPa, Yc = 250 MPa, S = 80 MPa.
This layer is subjected to a tension in the direction . Plot (first in Cartesian coordinates for 0 /2, then in polar coordinates for varying from 0 to 2) the tensile strength xu as a function of the angle , using the maximum stress criterion.
12.2 Do the preceding exercise again using Hoffmans criterion. Compare the results obtained. 12.3 The orthotropic layer of Exercise 12.1 is now subjected to a pure shear state in the direction . Plot (first in Cartesian coordinates, and then in polar coordinates) the shear strength xyu as a function of the shear angle , using the maximum stress criterion. 12.4 Do the preceding exercise again using Hoffmans criterion. Compare the results obtained.
Part IV
Mechanical Behaviour of Laminates and Sandwiches
This part develops the fundamental elements of the theory of laminate and sandwich plates. The general assumptions of the laminate theory are introduced first in Chapter 13. The classical theory of laminates is developed next in Chapter14. The study of the effect of the stacking sequence is analyzed in Chapter 15 and the analysis allows us to understand the phenomena of coupling between stretching, bending and twisting. Next, the classical laminate theory is applied for estimating the elastic behaviour of layers reinforced with cloths or mats. Chapter 16 establishes the fundamental equations of the classical laminate theory as well as the energy formulation. Accounting for the transverse shear effects in the theory of laminates is next developed in Chapter 17. Lastly, Chapter 18 presents the theory of sandwich plates which is based on the laminate theory which considers the transverse shear effects.
CHAPTER 13
Basics of Laminate Theory
13.1 INTRODUCTION 13.1.1 Architecture

In Chapter 3, we have considered the architecture of laminates which results from the designing of composite structures: as forms of plates and shells, as lamination of successive layers. This way of designing and manufacturing composite material structures justifies the importance of plates in the analysis of composite structures. In fact, besides plate types of structures, the analysis of plates also allows us to investtigate shell structures by using finite element analysis. In a general sense, a plate is a solid bounded by two parallel planes (Figure 13.1), the transverse dimension of which is small compared to the other two dimensions. It is then possible to define a reference plane between the two extreme planes, which is taken as the plane xy . The axis Oz corresponds to the transverse. z
y O
x
FIGURE 13.1. Plate element.
246
Chapter 13 Basics of Laminate Theory
13.1.2 Notations and Objective

The notations used are reported on Figure 13.2. The laminate is constituted of n layers, numbered from the lower to the upper face. The middle surface is chosen as the reference plane (Oxy) and the axis Oz is oriented in the direction of increasing number of the layers. Each layer k is referred to by the z coordinate of its lower face (hk-1) and upper face (hk). In Chapter 3 (Section 3.5), we have derived the process of analyzing the mechanical behaviour of a composite material structure. The process involves three steps: (i) the analysis of the mechanical behaviour at the level of the constituents, (ii) modelling the local mechanical behaviour of laminate or sandwich material, and then (iii) the analysis of the composite structure. Part III addresses the first step: analysis of the elastic behaviour (Chapters 9 to 11) and of the fracture behaviour (Chapter 12) of a layer. This chapter is the first of Part IV, the objective of which is to address the second step. The purpose of this fourth part is to establish modelling of the behaviour of laminate and sandwich plate, so as to simplify the analysis of the composite structure. It will be seen that this simplification consists in reducing the initial problem in three dimensions (x, y, z) of the mechanical behaviour of laminates and sandwiches to a less difficult analysis in two dimensions (x, y). The laminate theory introduces the same assumptions as the general theory of plates, assumptions that are developed in the present chapter.
13.2 DISPLACEMENT FIELD 13.2.1 General Expressions

The basic assumption of the laminate theory lies in expressing the displaycement at any point M, with coordinates (x, y, z), of a plate in the form of polynomial in z, usually limited to degree three, with coefficients dependent on (x, y). Thus, the displacement field is written in the form: z
layer number
n k
hk 1
middle plane
hk
h2 h 1 h 0
2 1 FIGURE 13.2. Laminate element.
13.2 Displacement Field
247
u ( x, y, z ) = u ( x, y, 0) + z x ( x, y ) + z 2 x ( x, y ) + z 3 x ( x, y ), v ( x, y, z ) = v ( x, y, 0) + z y ( x, y ) + z 2 y ( x, y ) + z 3 y ( x, y ), w ( x, y, z ) = w ( x, y, 0) + z z ( x, y ) + z 2 z ( x, y ).
(13.1)
This form of the displacement field satisfies the compatibility conditions for strains (6.18), and allows us to take into account a possible warping of a cross section of plates under deformation. In the case of dynamics analyses, the time factor must be introduced into Relations (13.1). The displacement of any point M (x, y, z) is thus developed, in accordance with (13.1), as a series in the variable z with coefficients in (x, y), starting from the reference point M0(x, y, 0) of the plane (Oxy). The displacement field at point M0 will be denoted by one of the notations:
u0 = u0 ( x, y ) = u ( x, y, 0), v0 = v0 ( x, y ) = v ( x, y, 0), w 0 = w 0 ( x, y ) = w ( x, y, 0).
(13.2)
13.2.2 Deformation of a Normal

Let us look for the deformation of line AB perpendicular to the plane of a plate, defined by ( x = a, y = b) (Figure 13.3). Every point M belonging to the normal AB has (a, b, z) as coordinates and its displacement is, by (13.1), written as:
u (a, b, z ) = u ( a, b, 0) + z x (a, b) + z 2 x (a, b) + z 3 x (a, b), v (a, b, z ) = v (a, b, 0) + z y (a, b) + z 2 y (a, b) + z 3 y (a, b), w (a, b, z ) = w (a, b, 0) + z z (a, b) + z 2 z (a, b).
The equation of the deformation of the normal AB is thus written, using obvious notations, in the polynomial for in z: u (a, b, z ) = Au + Bu z + Cu z 2 + Du z 3 ,
v (a, b, z ) = Av + Bv z + Cv z 2 + Dv z 3 , w (a, b, z ) = Aw + Bw z + Cw z 2 .
When the plate is deformed, the normal AB therefore suffers: a deformationless translation leading to A'B', consisting in a translation [ Au = u (a, b, 0), Av = v (a, b, 0)] in the plane (Oxy) and a translation [ Aw = w (a, b, 0)] along the axis Oz ; then a deformation leading to A"B", expressed in terms of z and the form of which depends of the degree of z. In the general case, the curve obtained is a warped curve. Thus, the displacement field (13.1) takes into account a possible warping of the normals when the plate is deformed.
248
A
Aw
A A
b
b + Av
M y
O a
a + Au
B
B
FIGURE 13.3. Deformation of line AB perpendicular to the middle plane.
13.2.3 First-Order Theory

The simplest and most widely used schemes (for example that of HenckyMindlin and of Kirchhoff) for describing the behaviour of plates reduce to firstorder theory of the form:
u ( x, y, z ) = u ( x, y, 0) + z x ( x, y ), v ( x, y, z ) = v ( x, y, 0) + z y ( x, y ), w ( x, y, z ) = w ( x, y, 0). or u ( x, y, z ) = u0 ( x, y ) + z x ( x, y ), v ( x, y, z ) = v0 ( x, y ) + z y ( x, y ), w ( x, y, z ) = w 0 ( x, y ). In a first-order scheme, the deformation of a normal AB is given by:
u (a, b, z ) = Au + Bu z , v (a, b, z ) = Av + Bv z , w (a, b, z ) = Aw .
(13.3)
(13.4)
In this case, the deformation A"B" remains a straight segment: the points located
13.2 Displacement Field
249
A H B
middle plane
A H
FIGURE 13.4. Deformation in the case of a first-order model without transverse shear.
on a normal to the middle plane (Oxy) before deformation stays on a straight segment after deformation. Furthermore, in the case where the transverse shear is not taken into account, (Subsection 14.1.1), the angles are not modified under deformation and the deformed segment of AB remains normal to the deformed middle plane (Figure 13.4). In this case, the deformation at point H (deformation of plane (Oxy) and deformation of the normal AB) is characterized (Figure 13.5) by: the displacements of point H : displacement in the plane (Oxy) [u(a, b, 0) = Au, v (a, b, 0) = Av] and transverse displacement [w (a, b, 0) = Aw] ; the rotations x and y about the directions i and j . In practice, it is more usual to characterize the rotation through the angles x and y (Figure 13.5), related to x and y by:
x = y
and
y = x .
(13.5)
First-order scheme allows us to solve the majority of usual problems. In the case where a first order scheme would not allow us to describe a given problem suitably, it would be necessary to consider a second-order, and even a third-order, scheme.
z
w
x
i
y y
H v
j
H x
FIGURE 13.5. Characterization of the deformation at a point when the transverse shear effect is not taken into account.
250
13.3 STRAIN FIELD 13.3.1 General Expressions

The strain field is expressed by Relations (8.19) in Cartesian coordinates. In the case of the general theory of third order, the strain field form is derived from Expressions (13.1) of the displacement field. Thus:
xx = yy
u u0 x = + z x + z2 + z3 x , x x x x x y y y v v 0 , = = +z + z2 + z3 y y y y y w = z ( x, y ) + 2 z z ( x, y ), z v w = 2 yz = + z y w z = y + 0 + z 2 y + z + z 2 3 y + , y y y u w = 2 xz = + z x z w = x + 0 + z 2 x + z + z 2 3 x + , x x x u v + y x 2 x y + + z x y 3 x y + + z x y .
zz = yz
(13.6)
xz
xy = 2 xy =
y u v = 0 + 0 + z x + x y x y
These expressions show that the truncation used in Expressions (13.1) of the displacement is consistent, in the sense that the transverse shear stress resulting from in-plane displacements are of the same order in z as the strains induced by the transverse displacement w.
13.3.2 First-Order Theory

In the case of a first-order scheme, the displacement field is expressed by Relations (13.4). The strain field is deduced simply from Relations (13.6) and is written as:
13.4 Strain Field
251
xx = yy
u u0 = +z x, x x x y v v0 , = = +z y y y w w 0 = = 0, z z v w w = 2 yz = + = y + 0 , z y y u w w + = x + 0 , z x x u v u0 v0 x y = + = + + . + z y x y x y x
zz = yz
(13.7)
xz = 2 xz = xy = 2 xy
This strain field is that of a first-order model which takes into account the transverse shear effects.
13.4 STRESS FIELD 13.4.1 General Expressions

The form of the stiffness matrix of a layer of unidirectional or woven fabric composite material, referred to the reference system (Oxyz) of the laminate, has been considered in Chapter 11. The stress state at point M of the laminate is expressed as a function of the strain field by Relation (11.3). If the point M belongs to the layer k of the laminate, the stress field is therefore written as: xx C11 C12 C13 0 C22 C23 0 yy C12 C23 C33 zz C13 0 = 0 0 0 C44 yz 0 0 C45 xz 0 C C C 0 26 36 xy 16 xx 0 C16 yy 0 C26 zz 0 C36 , C45 0 yz C55 0 xz 0 C66 k xy
(13.8)
are the stiffness constants of the layer k. where the components Cij In the general case, the strain field is given by Expressions (13.6). It results that the stresses in the layer k are polynomial in z. The plate theory simplifies the problem in three dimensions (x, y, z) in a problem in two dimensions (x, y). The reduction of the problem is obtained by integration of the stresses over the laminate thickness. This integration leads to the introduction of resultants and moments exerted over the plate. These resultants and moments will be defined in Section 13.5.
252
13.4.2
Simplification in the Context of the Plate Theory
The elementary theory of plates makes the assumption that the normal stresses zz are negligible within the volume of the plates, with respect to the other components xx, yy, xy. This assumption of a plane stress state is extended to the theory of laminates. That is: zz = 0. (13.9) This assumption is usually satisfied in practice. If that is not the case, the theory of plates can no longer be used. With the preceding assumption, Relation (13.8) relating stresses and strains is written: C12 C13 0 xx C11 C C C 0 22 23 yy 12 0 C23 C33 C13 0 = 0 0 C44 yz 0 xz 0 0 0 C45 C26 C36 0 xy k C16 xx 0 C16 0 C26 yy zz 0 C36 . 0 yz C45 0 xz C55 k 0 C66 xy
(13.10)
This relation can be rewritten by separating the transverse shear stresses and strains as follows:
C12 xx C11 C C 22 yy 12 C23 0 C13 = C26 xy C16 yz 0 0 0 xz k 0 C16 C13 C26 C23 C36 C33 C36 0 0 C66 0 0 0 0 0 0 C44 C45 0 0 0 0 C45 k C55 xx yy zz . xy yz xz
(13.11)
The state of the stresses xx, yy, xy and strains xx, yy, zz, xy corresponds to the plane stress state studied in Section 11.3. Applying the results obtained in this section, the stresses in the layer k are expressed by means of the reduced stiffness as follows: constants Qij
Q12 xx Q11 Q Q 22 yy 12 Q26 xy = Q16 0 yz 0 0 xz k 0 Q16 Q26 Q66 0 0 0 0 0 C44 C45 0 0 0 C45 C55 xx yy xy , yz k xz
(13.12)
with
13.5 Resultants and Moments
253
zz =
1 xx + C23 yy + C36 xy ) . ( C13 C33
(13.13)
of the reduced stiffness matrix of the layer k were introThe components Qij
k or Qij duced in (11.43). They will be denoted by Qij . The discontinuity of the stiffness matrix from one layer to another implies the discontinuity of the stresses when passing from one layer to another.
13.5 RESULTANTS AND MOMENTS 13.5.1 In-Plane Resultants

The field of the in-plane resultants, denoted N(x, y) in a matrix form, is defined as:
N ( x, y ) =
h2 h 2
k (M ) d z ,
(13.14)
where k(M) is the matrix of the in-plane stresses xx, yy, xy in the layer k. Thus: Nx xx h2 N ( x, y ) = N y = (13.15) yy d z . h 2 N xy xy
The components Nx, Ny, Nxy are the in-plane resultants, per unit length of the plate, respectively, of the normal stresses (in the directions x and y) and of the shear stress in the plane (x, y). They are represented symbolically in Figure 13.6. z
Nx
N xy
N xy
Ny
h N xy x
Nx
Ny N xy
FIGURE 13.6. In-plane resultants of the loads applied to a laminate element.
254
z y
dy dRxy
dRx
FIGURE 13.7. Resultants of the loading applied to a laminate surface element.
It is necessary to note that the resultants are relative to a unit length of cross section of the laminate. This means, for example, that the loading applied to a surface element normal to the direction x and of length dy (Figure 13.7) is the superposition of: dRx = N x dy, the normal resultant (13.16) dRxy = N xy dy. the shear resultant The discontinuity of the stresses from one layer to another leads to rewritting Relation (13.15) in the form:
Nx N ( x, y ) = N y = N xy
k =1
xx yy d z . hk 1 xy k
hk
(13.17)
13.5.2 Transverse Shear Resultants

The transverse shear resultants are defined in the same way by: Qx Q ( x, y ) = = Q y
k =1
xz dz . hk 1 yz k
hk
(13.18)
Like the in-plane resultants, the transverse shear resultants are loads per unit length of the laminate. They are illustrated in Figure (13.8).
13.5.3 Resultant Moments

The fundamental equations of laminates also introduce the resultant moments of the stresses applied to a laminate element. These moments are expressed by
13.5 Resultants and Moments
255
Qy x
Qx
Qx
Qy
FIGURE 13.8. Transverse shear resultants.
the expression: Mx M f ( x, y ) = M y = M xy
k =1
xx z yy d z . hk 1 xy k
hk
(13.19)
The components Mx and My are the bending moments in the directions x and y, respectively, and the component Mxy is the twisting moments. As previously, the resultant moments are moments per unit length. They are illustrated in Figure 13.9.
z Mxy Mxy Mxy h

Mx Mx
y My My
Mxy
x
FIGURE 13.9. Resultant moments applied to a laminate element.
256
13.6 FUNDAMENTAL EQUATIONS FOR PLATES IN THE CASE OF A FIRST-ORDER SCHEME 13.6.1 Fundamental Equations of the Mechanics of Materials
The fundamental equations of plates are derived from the fundamental relations (8.20), which we write in the form:
xx + xy + xz + f x = ax , x y z yy + yz + xy + f y = a y , y z x zz + xz + yz + f z = az , z x y
(13.20)
where fx, fy, fz are the components of the body forces acting at point M (x, y, z) of the material; ax, ay, az are the components of the acceleration vector at point M ; and is the density of the material at point M. The object of this section is to deduce from Relations (13.20) the fundamental equations for plates in the case of a first-order scheme.
13.6.2 Fundamental Equations for In-Plane Resultants

The integration of equations (13.20), through the thickness of the laminate, leads to the fundamental equations of a plate element for the resulants. The integration of the first two leads to the relations for in-plane resultants. For example, integration of the fist equation may be written:
xx dz + h 2 x
h2
h2 h 2
xy y
dz +
xz dz + h 2 z =
h2
h2 h 2
fx dz (13.21)
h2 h 2
a x d z.
The first term of this equation is:
Similarly:
xx dz = x h 2 x
h2
h2 h 2
xx d z =
N xy y
N x . x
(13.22)
h2 h 2
xy y
dz =
(13.23)
13.6 Fundamental Equations for Plates in the Case of a First-Order Scheme
257
The third term of Equation (13.21) is written as:
xz d z = xz ( h 2 ) xz ( h 2 ) , h 2 z
h2
where the stresses xz ( h 2 ) and xz ( h 2 ) are the shear stresses applied to the upper and lower faces of the laminate. These stresses are generally zero. In the case where it would be necessary to take then into account, we shall denote them:
xz ( h 2 ) = 1x
Whence:
and
xz ( h 2 ) = 2 x .
(13.24)
Lastly we set:
xz d z = 1x 2 x . h 2 x
h2
(13.25)
h2 h 2
f x d z = Fx .
(13.26)
The integration of the right-hand side of Equation (13.21) requires the expressions for the displacements as functions of x, y, z and time t. In the case of a first-order scheme, they are obtained by introducing the time into Expressions (13.4), thus:
u ( x, y, z , t ) = u0 ( x, y, t ) + z x ( x, y, t ), v ( x, y, z , t ) = v0 ( x, y, t ) + z y ( x, y, t ), w ( x, y, z, t ) = w 0 ( x, y, t ).
(13.27)
It results that:
Thus:
h2
h 2
ax d z =
=
h2
h 2
( x, y , z )
h2
2 u0
2 t
+z
2 x dz t 2
h2
2 u0 t 2
h 2
dz +
2 x t 2
h 2
z d z.
introducing:
h2
h 2
ax d z = s
2u0 t 2
+R
2 x t 2
(13.28)
s =
h2
h 2
dz
(13.29)
the weight per unit area of the laminate at point (x, y), and the quantity:
258
R=
h2
h 2
z dz .
(13.30)
The integration of the first equation of (13.20) thus leads to: N x N xy 2u 2 + + Fx + 1x 2 x = s 20 + R 2x . x y t t Similarly, the integration of the second equation of (13.20) yields:
N y y + N xy x + Fy + 1 y 2 y = s 2v0 t 2 +R 2 y t 2
(13.31)
(13.32)
where the components 1y and 2y take account of the possible shear stresses applied at the laminate faces:
yz ( h 2 ) = 1 y
and introducing the component:
Fy =
and
h2
yz ( h 2 ) = 2 y ,
(13.33)
h 2
f y dz .
(13.34)
13.6.3 Fundamental Equations for Transverse Shear Resultants

Integration of the third equation of (13.20) through the thickness of the laminate leads to:
zz dz + h 2 z
h2
h2
yz y
h 2
dz +
xz dz + h 2 x =
h2
h2
h 2
fz dz
h2
(13.35)
h 2
a z d z.
The second term is written as:
Similarly:
h2
yz y
h 2
dz =
h2
h2
h 2
yz d z =
Qy y
(13.36)

h2
xz Qx . dz = x h 2 x
(13.37)
The third term can be expressed in the form:

zz d z = zz ( h 2 ) zz ( h 2 ) , h 2 z
259
where the stresses zz ( h 2 ) and zz ( h 2 ) appear as components of the pressure forces applied to each face of the laminate. We note their difference as:
q ( x, y ) = q = zz ( h 2 ) zz ( h 2 ) .
(13.38)
Whence:
Lastly:
zz dz = q . h 2 z 2w 0 t 2
h2
(13.39)
to:
h2
h 2
az d z = s
(13.40)
Substitution of Equations (13.36) to (13.40) into Equation (13.35) finally leads Qx Q y 2w 0 + + q + Fz = s , x y t 2 introducing the component:
Fz =
h2
(13.41)
h 2
fz dz .
(13.42)
13.6.4 Fundamental Equations for Moments

The fundamental equations of moments are obtained multiplying the first two equations (13.20) by z, and then integrating through the thickness of the laminate. For example, the first equation leads:
M x M xy + + x y
h2
h 2
xz dz+ z
h2
h 2
zf x d z =
h2
h 2
zax d z. (13.43)
Integrating by parts, we obtain:
Thus:
h2 h 2
xz h2 d z = [ z xz ]h 2 z
h2 h 2
xz d z
h h = xz ( h 2 ) + xz ( h 2 ) Qx . 2 2
h2
h 2
xz h d z = (1x + 2 x ) Qx . z 2
(13.44)
260
The right-hand side of Equation (13.43) may be written:
Whence:
h2 h 2
zax d z =
=
h2 h 2
2u0
2 t
+z
2 x dz t 2 2 x t
2
2u0 t
2
h2 h 2
z dz +
h2 h 2
z 2 d z.
on setting:
h2
h 2
zax d z = R
2u0 t 2
+ I xy
2 x t 2
(13.45)
I xy =
h2
h 2
z2 dz .
(13.46)
The quantity Ixy is the moment of inertia about the middle plane (Oxy) of the plate element localized at point (x, y) and having sides of unit length. The first equation of the moments may therefore be written: M x M xy h 2u 2 x + + (1x + 2 x ) + Px Qx = R 20 + I xy , 2 x y t t 2 introducing the component of the moment of the body forces:
Px =
(13.47)
h2
h 2
zf x d z
Similarly, the second equation of (13.20) leads to:

M y 2 y h 2v0 + + (1 y + 2 y ) + Py Qy = R 2 + I xy , 2 y x t t 2 M xy
(13.48)
Introducing the component of moment:

Py =
h2
h 2
zf y d z .
(13.49)
13.6.5 Summary of Fundamental Equations

The fundamental equations of plates thus consist of Equations (13.31), (13.32), (13.41), (13.47) and (13.48). By regrouping them, we obtain:
261
N x N xy 2u0 2 x + + Fx + 1x 2 x = s 2 + R 2 , x y t t
N y y + N xy x Q y + Fy + 1 y 2 y = s 2v0 t 2 +R 2 y t 2
, (13.50)
Qx 2w 0 , + + q + Fz = s x y t 2 M x M xy h 2u 2 x + + (1x + 2 x ) + Px Qx = R 20 + I xy , 2 x y t t 2
M y 2 y h 2v0 + + (1 y + 2 y ) + Py Qy = R 2 + I xy , 2 y x t t 2 M xy
with
s , R, I xy =
(1, z, z ) d z .
2
h2
h 2
The last three equations of (13.50) allow us to obtain an equation which is independent on the transverse shear resultants, as follows: 2M x x 2 + 2M y y 2 +2 2 M xy xy +q (13.51)
= s
2w 0 t 2
3 x 3 y 3u0 3v0 . + R + +I + xt 2 yt 2 xy xt 2 yt 2
Equations (13.50) and (13.51) constitute the equations of motion for the usual theory of plates. They are applicable to homogeneous plates as well as to laminated plates. The first two equations of (13.50), associated with (13.51), constitute the fundamental equations of plates when the transverse shear effects are neglected. They are: N x N xy 2u 2 + + Fx + 1x 2 x = s 20 + R 2x , x y t t
N y y
2
N xy x
2
+ Fy + 1 y 2 y = s
2v0 t 2
+R
2 y t 2
(13.52)
Mx x 2
My y 2
+2
M xy xy
+q
= s
2w 0 t 2
3 3u0 3 y 3v0 x . + R + +I + xt 2 yt 2 xy xt 2 yt 2
262
The quantities s, R and Ixy are evaluated easily in the case where the plate is constituted of n layers, the layer k having a density k. We have, according to (13.29) :
s =
h2 h 2
dz =
h
k =1 n
hk
k 1
k d z =
k ( hk hk 1 ) .
k =1
(13.53)
Similarly, the quantities R and Ixy are expressed as: R= 1 2 2 k hk hk 1 , 2 k =1 1 3 3 k hk hk 1 . 3 k =1
(
n
(13.54)
I xy =
(13.55)
In the most cases, the rotational inertia terms can be neglected, and in the absence of body forces and shear stresses on the plate faces, the equation of plates simplify as follows:
N x N xy 2 u0 + = s 2 , x y t N y y + N xy x = s 2v0 t 2 ,
Qx Qy 2w 0 + + q = s , x y t 2 M x M xy + Qx = 0, x y M y y + M xy x Qy = 0.
(13.56)
These equations can also be written eliminating the transverse shear resultants in a form analogous to (13.52). Whence: N x N xy 2u + = s 20 , x y t N y y 2M x x 2 + 2M y y 2 +2 + N xy x = s 2v0 t 2 2w 0 t 2 , . (13.57)
2 M xy xy
+ q = s
263
Exercises
13.6.6 Statics Problems

In the case of statics problems, the displacements are independent of time and the fundamental equations of plates reduce to:
N x N xy + = 0, x y N y y + N xy x = 0,
Qx Qy + + q = 0, x y M x M xy + Qx = 0, x y M y y + M xy x Qy = 0.
(13.58)
Or eliminating the transverse shear resultants, we have:

N x N xy + = 0, x y N y y 2M x x 2 + 2M y y 2 +2 + N xy x = 0,
(13.59)
2 M xy xy
+ q = 0.
EXERCISES
13.1 A laminate is constituted of three layers 1, 2 and 3, respectively oriented in the directions 0, 30 and 45. These layers, each of the same thickness h = 1 mm, have the same engineering constants:
EL = 160 GPa, ET = 15 GPa,
LT = 0.32,
GLT = 5 GPa,
GTT = 4.5 GPa.
At a point, the laminate is subjected to the strain state:
264
Exercises
xx = 0.40 %, yy = 0.25 %, xy = 0.50 %, xz = yz = 0.50 %.

Calculate the stresses in each layer; then the in-plane and shear resultants, the resultant moments to obtain this strain state.
13.2 Do Exercise 13.1 again in the case of the order of the layers of the laminate is reversed. Compare the results obtained in the two exercises.
CHAPITRE 14
Classical Laminate Theory
14.1 STRAIN FIELD 14.1.1 Assumptions of the Classical Laminate Theory

The classical theory of laminates uses a first-order scheme for the strains (13.7). Next, the theory makes an additional assumption that consists of neglecting the transverse shear effects. In this scheme, the transverse shear strains are zero, thus:
xz = 0
and
yz = 0 .
w 0 , x
(14.1)
This assumption implies, from (13.7):
x ( x, y ) =
w y ( x, y ) = 0 . y The displacement field is then, by (13.4), written as:
(14.2)
u ( x, y, z ) = u0 ( x, y ) z v ( x, y , z ) = v 0 ( x, y ) z w ( x, y, z ) = w 0 ( x, y ).
w 0 ( x, y ), x w 0 ( x, y ), y (14.3)
The deformation of the normal to the middle plane (Oxy) is then a straight segment normal to the deformed middle plane (Subsection 13.2.3 and Figure 13.4). The deformations and the notations, used in the case of the classical laminate theory, are illustrated in Figure 14.1.
266
Chapter 14 Classical Laminate Theory
x
z A M H B
v0
z x
y
z A M H B
u0
z y
A M H y B
x
w0
A M H x B
y
w0
FIGURE 14.1. Representation of plate deformation in the case of the classical laminate theory.
14.1.2 Expression for the Strain Field

From (13.7) and taking account of Expressions (14.2), the strain field is written:
xx yy
u0 2w 0 = z , x x 2 v0 2w 0 = z , y y 2 (14.4)
zz = 0, yz = 0, xy xz = 0,
u0 v0 2w 0 = + . 2z x xy y
14.1 Stain Field
267
The strain tensor at point M is:

xx ( M ) = xy 0
xy yy
0
0 , 0 0
(14.5)
and the strain matrix reduces to three nonzero components:

xx ( M ) = yy . xy
(14.6)
The strain field is the superposition of: the in-plane strains (or midplane strains):
u0 0 xx x v 0 0 = m ( M ) = yy , y 0 u v xy 0 + 0 x y
(14.7)
being expressed solely as functions of the displacements (u0, v0) in the plane (Oxy) of the points of this plane; the flexural strains (bending and twisting strains):
2w 0 z f x 2 xx 2w 0 , f ( M ) = f yy = z y 2 f 2 xy 2 z w 0 xy
(14.8)
being expressed as functions of the rotation angles of the deformed midplane and of the z coordinate of point M. Usually, the bending and twisting strains are expressed by the relation:
f ( M ) = z ( x, y ) ,
(14.9)
268
on introducing :
2w 0 x 2 x 2 w0 . ( x, y ) = y = 2 y xy 2 2 w 0 xy
(14.10)
The matrix ( x, y ) is called the curvature matrix of the plate subjected to bending and twisting. The rotation angles of the deformed midplane at the point H(x, y, 0) are expressed (Figure 14.1) as functions of the transverse displacement w0(x, y) at this point by: w x = 0 in the direction i , y (14.11) w 0 y = in the direction j . x
The displacement field (14.3) is then written:

u ( x, y, z ) = u0 ( x, y ) z y ,
v ( x, y, z ) = v0 ( x, y ) z x , w ( x, y, z ) = w 0 ( x, y ).
(14.12)
Finally, the strain field is expressed as:

( M ) = m ( M ) + f ( M ) ,
(14.13)
or
0 xx xx x 0 yy = yy + z y , 0 xy xy xy
(14.14)
with
0 = xx
u0 ( x, y ), x 2w 0 x 2
0 yy =
v0 ( x, y ), y 2w 0 y 2
0 = xy
u0 v0 + , y x
x =
( x, y ),
y =
( x, y ),
xy
2w 0 = 2 ( x, y ). xy
(14.15)
14.2 Stress Field
269
In condensed form, the strain field is then written:

( M ) = ( x , y , z ) = m ( x, y ) + z ( x, y ) .
(14.16)
The in-plane strains m(x, y) and curvatures (x, y) depend only on the coordinates (x, y) of point H of the midplane of the laminate.
14.2 STRESS FIELD 14.2.1 Form of the Stress Field

The stress field is deduced from Relation (13.12). In the case of the classical laminate theory, we obtain for the layer k:
xx + Q12 yy + Q16 xy , xx = Q11 xx + Q22 yy + Q26 xy , yy = Q12 xx + Q26 yy + Q66 xy , xy = Q16 (14.17)
yz = 0, xz = 0.
The stress tensor at point M is thus of the form:
xx xy (M ) = xy yy 0 0 0 0 . 0
(14.18)
The stress field reduces to the in-plane stresses: xx, yy and xy.
14.2.2 Stress Expression

Relations (14.17) show that the stresses in layer k are expressed as: xx yy = Q k xy k with xx yy , xy
(14.19)
270
Q12 Q11 Q22 Q k = Q12 Q Q 26 16
Q16 Q26 Q66 k
where Q k is the reduced stiffness matrix of the layer k introduced in (11.43) and the components of which are expressed as functions of the engineering constants by Relations (11.52). On taking (14.14) into account, the stresses in the layer k are expressed as follows:
xx Q11 Q12 Q22 yy = Q12 Q Q26 xy 16 k Q16 Q26 Q66 k
0 xx Q11 Q12 0 Q22 yy + z Q12 0 Q Q26 xy 16
Q16 Q26 Q66 k
x y , (14.20) xy
or
k ( M ) = k ( x , y , z ) = Q k m ( x, y ) + z Q k ( x, y ) .
(14.21)
The matrix k ( M ) represents the stress matrix in the layer k: hk1 z hk. The reduced stiffness matrix Q k changes from one layer to another. From this it results that there is a discontinuity in the in-plane stress field between successive layers.
14.3 EXPRESSIONS OF RESULTANTS AND MOMENTS 14.3.1 In-Plane Resultants

Expression (13.17) associated with Relation (14.20) or (14.21) leads to the expression for the in-plane resultants, in the context of the classical laminate theory. We obtain:
N ( x, y ) = Whence:
h
k =1
hk
k 1
[Qk m ( x, y) + z Qk ( x, y)] d z .
N ( x, y ) =
Q k m ( x, y ) k =1
hk hk 1
d z + Q k ( x, y )
z dz , hk 1
hk
or integrating through the thickness of the laminate:

n n 1 2 2 + N( x, y ) = ( hk hk 1 ) Q x y h h Q ( , ) k m k k 1 k ( x, y ) . 2 k =1 k =1
14.3 Expression of Resultants and Moments
271
The preceding expression of the matrix of the in-plane resultants can be expressed finally in the form:
N ( x, y ) = A m ( x , y ) + B ( x , y ) ,
(14.22)
by introducing the matrices: A=
( hk hk 1 ) Qk ,
k =1
A= Aij and B=
n
with
Aij =
( hk hk 1 ) ( Qij )k ,
k =1
(14.23)
2 ( hk2 hk21 ) Qk ,
1
k =1
B= Bij
with
1 2 2 ) . hk Bij = hk 1 ( Qij k 2 k =1
(14.24)
The extended expression for the in-plane resultants is therefore written as: N x A11 N y = A12 N xy A16 A12 A22 A26
0 A16 xx B11 0 A26 yy + B12 0 A66 B16 xy
B12 B22 B26
B16 x B26 y . B66 xy
(14.25)
This expression shows that, in the case of a laminate, the in-plane resultants (Nx,
0 0 0 Ny, Nxy) are not functions only of the in-plane strains xx , yy , xy (as in the case of homogeneous plates), but are also functions of the bending and twisting curvatures (x, y, xy).
14.3.2 Resultant Moments

The resultant moments are obtained by introducing Expression (14.21) for the stresses into Expression (13.19). Hence: M f ( x, y ) = which yields:
1 n 2 1 n 3 3 2 M f ( x, y ) = hk hk 1 Q k m ( x, y ) + hk hk 1 Q k ( x, y ) . 3 k =1 2 k =1
h
k =1
hk
k 1
z Q ( x , y ) + z 2 Q k ( x, y ) d z , k m
272
The matrix of the flexural moments is thus written as:

M f ( x, y ) = B m ( x , y ) + D ( x , y ) ,
(14.26)
on introducing the new matrix: D=
3 ( hk3 hk31 ) Qk ,
1
k =1
D= Dij
with
1 3 3 ) . hk Dij = hk 1 ( Qij k 3 k =1
(14.27)
The extended expression of the moments can thus be written in the form: M x B11 M y = B12 M xy B16 B12 B22 B26
0 B16 xx D11 0 B26 yy + D12 0 B66 xy D16
D12 D22 D26
D16 x D26 y . D66 xy
(14.28)
The bending and twisting moments are therefore functions of the bending and twisting curvatures, but are also functions of the in-plane strains.
14.4 MECHANICAL BEHAVIOUR EQUATION OF A LAMINATE 14.4.1 Constitutive Equation

The constitutive equation of a laminated plate expresses the resultants and the moments as functions of the in-plane strains and of the curvatures. It is obtained by regrouping Expressions (14.25) and (14.28) into a single matrix equation of the form: N x A11 N y A12 N xy = A16 M x B11 M y B12 M xy B16 A12 A22 A26 B12 B22 B26 A16 A26 A66 B16 B26 B66 B11 B12 B16 D11 D12 D16 B12 B22 B26 D12 D22 D26
0 B16 xx B26 0 yy 0 B66 xy . D16 x D26 y D66 xy
(14.29)
The constitutive equation can be also expressed in a condensed form as follows:
14.4 Mechanical Behaviour Equation of a Laminate
273
N A B m = . M B D
(14.30)
The terms of the matrices A, B and D are given by Expressions (14.23), (14.24) and (14.27). They can also be expressed by introducing the thickness ek and the z coordinate zk of the middle plane of layer k, in the form: Aij =
(Qij )k ek ,
k =1 n
(14.31)
Bij =
n
(Qij )k ek zk ,
k =1
(14.32)
3
Dij =
e e z2 + k . (Qij )k k k 12
k =1
(14.33)
The coefficients Aij, Bij, Dij of the constitutive equation (14.29) of a laminate are thus expressed as functions of the reduced stiffness constants of the layers. For each layer, the reduced stiffness constants are expressed as functions of the engineering constants from Relations (11.52) and Expressions reported in Table 11.6.
14.4.2 Stiffness Matrix of a Laminate

The matrix introduced in Expression (14.29) is the stiffness matrix of the laminate, which describes the elastic behaviour of the laminate at point M0(x, y) = M(x, y, 0). Matrix A is called the in-plane stiffness matrix, D is the flexural stiffness matrix and B is the coupling matrix between in-plane and flexural behaviours of the laminate. This coupling is induced all the same as the materials of the layers are homogeneous. It results from the structure in the form of layers with mechanical characteristics which are different. The coupling matrix vanishes (B = 0), only when the laminate is symmetric (Chapter 15). Different couplings can be observed and illustrated (Chapter 15). The coupling between in-plane tension and in-plane shear is induced by the terms A16 and A26. The coupling between in-plane and bending behaviours results from the terms B11, B12 and B22, when the coupling between in-plane and twisting behaviours is introduced by the terms B16 and B26. Lastly, the coupling between bending and twisting results from the coefficients D16 and D26. Different types of laminates will be considered in Chapter 15.
274
14.4.3 Examples
14.4.3.1 Example 1
We consider a laminate constituted of two unidirectional layers (Figure 14.2). The lower layer has a thickness of 3 mm and is oriented at 45 from the reference system (x, y, z) of the laminate. The upper layer is oriented at 0 and is 5 mm thick. The unidirectional composite material of the two layers is a glass fibreepoxy composite with engineering constants:
EL = 46 GPa, ET = 10 GPa, GLT = 4.6 GPa, LT = 0.31.
We explicit the constitutive equation of the laminate. 1. Determination of the reduced stiffness constants in the material directions
Q11 = EL
2 1 LT
ET EL
= 46.982 GPa,
Q22 =
ET Q11 = 10.213 GPa, EL
Q12 = LT Q22 = 3.166 GPa, Q66 = GLT = 4.6 GPa.
2. Stiffness matrices of layers in the laminate directions Layer oriented at 0 46.982 3.166 0 0 GPa . Q 0 = 3.166 10.213 0 0 4.6
z y
2 1
5 mm 3 mm
x
FIGURE 14.2. The laminate with two layers considered in Example 1.
275
Layer oriented at 45
= ( Q11 + Q22 + 2Q12 + 4Q66 ) cos 4 45 = 20.482 GPa, Q11 = ( Q11 + Q22 4Q66 + 2Q12 ) cos 4 45 = 11.282 GPa, Q12 = ( Q11 Q22 ) cos 4 45 = 9.192 GPa, Q16 = Q11 , Q22 = Q16 , Q26 = ( Q11 + Q22 2Q12 ) cos 4 45 = 12.716 GPa. Q66 Whence: 20.482 11.282 9.192 = 11.282 20.482 9.192 GPa . 9.192 9.192 12.716
Q 45
3. Matrix A
Aij =
(Qij )k ek
k =1
) + 5 ( Qij ) 103. = 3 ( Qij 45 0 Hence: 296.35 49.676 27.576 A = 49.676 112.51 27.576 106 Nm-1 . 27.576 27.576 61.147
4. Matrix B
1 2 2 ) Bij = hk hk 1 ( Qij k 2 k =1 ) + ( Qij ) 106. = 7,5 ( Qij 45 0 Whence: 198.75 60.87 68.94 B = 60.87 77.01 68.94 103 N . 68.94 68.94 60.87
276
5. Matrix D
Dij = 1 3 3 ) hk hk 1 ( Qij k 3 k =1
65 ) + ( Qij ) = 21( Qij 109. 45 0 3 Thus: 1448.07 305.52 193.03 D = 305.52 651.40 193.03 Nm . 193.03 193.03 366.70
6. Constitutive equation of the laminate

Combining the preceding results, the constitutive equation of the laminate can be written as :
Nx Ny N xy = Mx My M xy 296.35 106 49.676 106 6 27.576 10 3 198.75 10 60.87 103 3 68.94 10 49.676 106 112.51 106 27.576 106 60.87 103 77.01 10
3 3
27.576 106 27.576 106 61.147 106 68.94 103 68.94 10

3 3
198.75 103 60.87 103 68.94 103 1448.07 305.52 193.03
60.87 103 77.01 103 68.94 103 305.52 651.40 193.03
68.94 10
60.87 10
68.94 103 xx 0 68.94 103 yy 0 60.87 103 xy 193.03 x 193.03 y 366.70 xy

0
14.4.3.2 Example 2
We shall now consider the laminate of Figure 14.3, constituted of four unidirectional layers with the same properties:
The thicknesses and the orientations of the layers are reported in Figure 14.3. This laminate is antisymmetric (Chapter 15): the thicknesses of the layers are symmetric, the orientations of the layer are antisymmetric.
277
1 mm 1.5 mm 1.5 mm 1 mm
= 30
= 15
4 3
= 15 = 30
2 1
FIGURE 14.3. Four-layer laminate of Example 2.
1. Reduced stiffness constants in the material directions

Q11 = 38.945 GPa, Q22 = 9.224 GPa, Q12 = 2.952 GPa, Q66 = 3.6 GPa, Q16 = 0, Q26 = 0.
2. Stiffness matrices of layers in the laminate directions

Layer 1 oriented at 30 26.290 8.176 9.451 = 8.176 11.429 3.418 GPa . 9.451 3.418 8.825
Q 30
Q15
Q 15
278
Layer 4 oriented at 30
Q 30
26.290 8.176 9.451 = 8.176 11.429 3.418 GPa . 9.451 3.418 8.825
3. Matrices A, B and D
) Aij = ( Qij
30
) + ( Qij
30
) + ( Qij ) . + 1.5 ( Qij 15 15
Whence: 0 158.22 30.432 A = 30.432 51.277 0 106 Nm 1 . 0 0 33.674 Furthermore : ) ( Qij ) + 1.125 ( Qij ) ( Qij ) , Bij = 2 ( Qij 30 30 15 15 Dij = 1 ) + ( Qij ) + 3.375 ( Qij ) + ( Qij ) . 12.25 ( Qij 30 15 30 15 3
Hence the matrices B and D: 0 22.659 0 B= 0 0 12.101 103 N , 22.659 12.101 0 0 293.93 77.332 0 Nm . D = 77.332 114.65 0 0 84.087
4. Stiffness matrix of the laminate

Combining the preceding results the stiffness matrix of the laminate is:
158.22 106 30.432 106 0 0 0 3 22.659 10
30.432 106 51.277 106 0 0 0 12.101 10
3
0 0 33.676 10 12.101 10 0
6
0 0 22.659 10 293.93 77.332 0

3
0 0 12.101 103 77.332 114.65 0
22.659 103
3
22.659 103 12.101 103 0 . 0 0 84.087
279
14.4.3.3 Example 3
The effect of the stacking sequence of the layer can be illustrated by considering the laminate of Figure 14.4, obtained by reversing the layers 1 and 2, oriented at 15 and 30 of the laminate (Figure 14.3) of the preceding example. The elements of the matrix A are expressed as:
) Aij = ( Qij
30
) + ( Qij
30
) + ( Qij ) . + 1.5 ( Qij 15 15
The matrix A of the sequence [15/30/15/30] stays unchanged with respect of the preceding sequence [30/15/15/30]. The matrices B and D are easily determined. We obtain: 13.384 5.2247 1.6154 B = 5.2247 2.9342 5.9258 103 N , 1.6154 5.9258 5.2247 327.38 64.271 60.686 D = 64.271 107.32 15.438 Nm . 60.686 15.438 71.025 The change of the stacking sequence of the layers keeps the in-plane matrix unchanged, when it modifies the flexural and coupling matrices.
1 mm 1.5 mm 1 mm 1.5 mm
= 30
= 15 = 30 = 15
4 3 2 1
FIGURE 14.4. Four-layer laminate of Example 3.
280
14.5 DETERMINATION OF STRAINS AND STRESSES 14.5.1 The Problem to Be Solved

The constitutive equation (14.29) expresses the in-plane resultants Nx, Ny, Nxy and the bending-twisting moments Mx, My, Mxy as functions of in-plane strains
0 0 xx , 0 yy , xy , and of the curvatures x, y, xy. The problems of designing
structures of composite materials require the solution of the inverse problem: knowing the in-plane resultants and the moments, find the in-plane strains and the curvatures, and then the stresses.
14.5.2 In-Plane Strains and Curvatures

Expression (14.30) can be written separating the matrix N of the resultants and the matrix Mf of the moments:
N = A m + B , Mf = B m + D .
(14.34) (14.35)
From the first equation, we can solve for the in-plane strains. Thus:
m = A 1N A 1B ,
(14.36)
and substituting in (14.35), the matrix of the moments is written:

M f = B A 1N + (D B A 1B) .
(14.37)
Expressions (14.36) and (14.37) can be rewritten in a partially inverted form:

m = A N + B , M f = CN + D .
(14.38) (14.39)
Whence, the matrix form:

m A = M f C B N D ,
(14.40)
introducing the matrices A , B , C and D , such as:
14.5 Determination of Strains and Stresses
281
A = A 1 , B = A 1B, C = BA
1 t
(14.41)
= B ,
D = D B A 1B = D + B B . In the general case, the matrices A* and D* are symmetric, when the matrix B* is not. From Expression (14.39), we obtain:
= D1M f D1CN ,
(14.42)
and substituting in (14.38): m = ( A BD1C ) N + BD1M f . (14.43)
Equations (14.42) and (14.43) can be regrouped so as to obtain the fully inverted form of the constitutive equation of laminates. Thus: m A = C with B N , D M f (14.44)
A = A BD1C = A + BD1Bt , B = BD1 , C = D

1
C = B ,
(14.45)
D = D1.
The inverted form of the constitutive equation is written in (14.44) in a form analogous to that of (14.30). It introduces inversions of 3 3 submatrices of the direct form. The inverted stiffness matrix can also be obtained by direct inversion of the 6 6 stiffness matrix in equation (14.29).
14.5.3 Strain Field

The strain field at point (x, y, z) is next determined from the in-plane strains and curvatures considering Expression (14.14). Taking account of the assumptions made (first-order scheme), the in-plane strains xx, yy and xy vary linearly through the thickness of the laminate. The strains, expressed in the material directions of the layer with orientation with respect to the reference system of the laminate (Figure 11.1), are next
282
deduced from the general transformation relation (6.41). In the present case, the transformation relation is restricted to the three in-plane strains. Thus, the strains in the layer k, referred to the material directions (L, T, T') of the layer, are written in the form:
xx L T = T yy , LT k xy
(14.46)
where the transformation matrix is expressed as:
cos 2 T = sin 2 2sin cos
sin 2 cos 2 2sin cos
sin cos . cos 2 sin 2

sin cos
(14.47)
14.5.4 Stress Field

The stresses in the layer k are next obtained using Equation (14.19) or (14.20). For example :
Q12 xx Q11 Q22 yy = Q12 Q 26 xy k Q16
Q16 Q26 Q66 k
xx yy . xy
(14.48)
The stresses, expressed in the material directions of the layer, are next derived either by using the stress transformation on the in-plane stresses xx, yy and xy, or directly from the in-plane strains L, T, LT in the material directions. By applying the transformation equation of the stresses, the expression is obtained from the general relation (5.44). We obtain:
xx L T = T yy , LT k xy k
introducing the matrix:
(14.49)
cos 2 t T = [ T( )] = sin 2 sin cos
sin 2 cos 2 sin cos
2sin cos 2sin cos . cos 2 sin 2
(14.50)
283
Starting from the in-plane strains expressed in the material directions, the stresses in the layer k are obtained as:
L Q11 Q12 T = Q12 Q22 0 LT k 0
0 0 Q66 k
L T . LT k
(14.51)
14.5.5 Example
The loads applied to a structure constituted of a laminated material are such that they reduce at a point to the in-plane resultants Nx, Ny, Nxy (Figure 14.5). We must determine at the point under consideration: 1. the in-plane strains and the curvatures; 2. the strains in each layer referred to the reference system (x, y, z) of the laminate, then to the material directions; 3. the stresses in each layer referred to the reference system (x, y, z) of the laminate, then to the material directions; in the case where the laminate is that of example 3 of Subsection 14.4.3.3 (Figure 14.4), and the values of the in-plane resultants are :
N x = 1,000 N/mm, N y = 500 N/mm, N xy = 250 N/mm.
y x
1 mm
Ny Nxy Nx
1 mm
FIGURE 14.5. In-plane loads acting at a point of the laminate.
284
1. In-plane strains and curvatures
Taking into account the results established in Example 3 of Subsection 14.4.3.3, the constitutive equation of the laminate is written as:
1000 500 250 103 = 0 0 0 158.22 106 30.432 106 0 3 13.384 10 5.2247 103 3 1.6154 10 30.432 106 51.277 106 0 2.9342 103 5.9258 103 0 0 33.674 106 5.9258 103 5.2247 103 13.384 103 5.2247 103 1.6154 103 xx 0 5.2247 103 2.9342 103 5.9258 103 yy 0 1.6154 103 5.9258 103 5.2247 103 xy . 327.38 64.271 60.686 x 64.271 107.32 15.438 y 60.686 15.438 71.025 xy
0
5.2247 103 1.6154 103
By direct inversion or using Equation (14.44), the inverted equation is obtained as:
0 xx 0 yy 0 xy = x y xy
7.207 4.322 0.069 22.297 0.279 4.322 0.069 0.279 30.508 3 3 3 0.415 10 0.187 10 1.032 10 0.525 103 0.042 103 1.920 103 0.279 103 1.810 103 2.730 103
0.415 103 1.032 103
0.525 103 1.920 103
0.187 103 0.042 103 4.052 106 2.058 106 2.058 106 10.747 106 3.065 106 0.445 106
0.279 103 1 1.810 103 0.5 2.730 103 0.25 103. 3.065 106 0 0.445 106 0 17.15 106 0
Whence the in-plane strains and the curvatures:

0 xx = 5.064 103 ,
x = 0.,580, y = 1.027, xy = 1.309.
0 yy 0 xy
= 6.897 10 , = 7.836 10 ,
3
In the above inverted equation, it is to be noted that the inverted matrix is symmetric, as the stiffness matrix of the constitutive equation. It is the same for
285
the submatrices A' and D'. In contrast, the submatrices C' and B' (transposed from each other) are not symmetric.
2. Strains in the layers
The strains, referred to the reference system (x, y) of the laminate, are derived from Relation (14.14) and are expressed as follows:
xx 5.064 0.580 3 xy = 6.897 10 + 1.027 z . 1.309 xy 7.836
The variations of the in-plane strains xx, yy and xy as functions of z are reported in Figure 14.6. The strains in each layer, referred to the material directions of the layer, are next deduced from Relation (14.46). For the layer k of the laminate, we have:
L T = Ak + Bk z , LT k
(14.52)
with
0 xx A k = T 0 yy , 0 xy
x Bk = T y , xy
(14.53)
where T is the transformation matrix defined in (14.47). From that we deduce: L 8.914 0.388 3 T = 3.045 10 + 0.058 z , 2.046 LT 30 5.505
L 2.129 0.745 = 9.831 103 + 1.192 z , T 0.737 LT 30 2.330 L 7.145 0.145 3 T = 4.815 10 + 0.592 z , 1.937 LT 15 7.702
1.5 mm z 2.5 mm,
1 mm z 0,
2.5 mm z 1 mm,
286
L 3.227 0.799 = 8.733 103 + 1.246 z, T 0.330 LT 15 5.869
0 z 1.5 mm.
Whence the relations giving the strains L, T, LT as functions of the z coordinate:

2.5 mm z 1 mm 1 mm z 0
L = 7.145 103 + 0.145 z, T = 4.815 103 0.592 z, LT = 7.702 103 1.936 z.

0 z 1.5 mm
L = 2.129 103 + 0.745 z , T = 9.831 103 1.191z, LT = 2.330 103 + 0.737 z.

1.5 mm z 2.5 mm
L = 3.227 103 + 0.799 z , T = 8.733 103 1.246 z, LT = 5.869 103 0.330 z.
L = 8.914 103 + 0.388 z, T = 3.046 103 0.058 z, LT = 5.505 103 2.046 z.
The variations of the strains L, T, LT through the thickness of the laminate are reported in Figure 14.6.
3. Stresses in the layers
The stresses in each layer, referred to the reference system (x, y) of the laminate, are written from Relations (14.19) or (14.20) as follows: xx yy = A1k + B1k z , xy k with
0 xx 0 = Q k yy , 0 xy
(14.54)
A1k
x B1k = Q k y , xy
(14.55)
where the matrices Q k are the reduced stiffness of the layers determined in Example 2 of Subsection 14.4.3.2. From this we deduce:
287
6.5 30 15 30 15
4.3
4.6
xx 3.6
7.9 30 15 30 15 2.1 3.2 7.0 1.4 4.4 8.3
yy
2.9 6.9 2.96
9.5 0.4
xy
( 103 )
11.1
2.4 9.8 8.7 2.3 1.5
5.3
5.9 9.6
5.4
11
6.8
T
250
6.3
LT
118 74 12 130 93 102 2 4 85 11
( 103 )
12.5 116 126
30 15 115 30 96 15 158
194 256
87 257
96
xx
246 318
yy
114 50 52 76 90
xy
1.4 19 9 97
( MPa )
91
30 15 112 30 86 15 151
193
333
289
71
106
8 21 35 6
283
78
LT
45 ( MPa )
FIGURE 14.6. Stresses and strains through the thickness of the laminate.
288
xx 263.564 5522.83 yy = 147.009 + 11467.79 z, (MPa) xy 30 140.579 9580.17 xx 115.451 19217.67 = 93.440 + 2519.78 z, (MPa) yy xy 30 2.284 13520.26 xx 263.411 6785.75 yy = 94.571 + 7919.24 z , xy 15 80.758 3805.83
1.5 mm z 2.5 mm,
1 mm z 0,
(MPa)
2.5 mm z 1 mm,
xx 157.918 24407.03 = 83.622 + 6090.45 z, (MPa) yy xy 15 2.951 10176.80
0 z 1.5 mm.
The variations of the stresses xx, yy and xy through the thickness of the laminate are reported in Figure 14.6. The stresses in each layer, referred to the material directions of the layer considered, are next deduced from Relation (14.49). Hence: L T = A 2k + B 2k z , LT k with
A 2 k = TA1k , B 2 k = TB1k ,
(14.56)
(14.57)
where T' is the transformation matrix for the stresses defined in (14.50). We obtain :
L 356.170 15305.74 T = 54.402 + 1684.88 z , (MPa) LT 30 19.820 7364.33 L 111.926 25492.20 = 96.965 + 8794.31 z , (MPa) T LT 30 8.389 2652.46
1.5 mm z 2.5 mm,
1 mm z 0,
Exercises
289
L 292.480 3987.79 T = 65.502 + 5031.28 z, LT 15 27.728 6972.19
(MPa)
2.5 mm z 1 mm,
L 151.466 27452.49 = 90.074 + 9135.91 z , (MPa) T LT 15 21.130 1189.00
0 z 1.5 mm.
The variations of the stresses L, T and LT through the thickness of the laminate are reported in Figure 14.6. These variations allows us to evaluate the conditions of first fracture in the laminate, by applying to each layer the fracture criteria considered in Chapter 12.
EXERCICES
14.1 A [0/30/45] laminate is constituted of three layers of the same thickness e = 1 mm and of the same mechanical characteristics:
Calculate the stiffness matrix of the laminate.

14.2 The layers considered in Exercise 14.1, of thicknesses equal to 0.5 mm, now constitute a symmetric [0/30/45]s laminate. Calculate the new stiffness matrix. Compare it with the previous matrix. 14.3 Do Exercises 14.1 and 14.2, reversing the order of the layers: [45/30/0] and [45/30/0]s . Compare the different results obtained. 14.4 Do exercises 14.1 and 14.2 again, modifying the orientation of the layers to [0/45/90]. 14.5 Implement a numerical procedure having :
as inputs : the number n of layers, the moduli EL, ET, LT GLT and the orientation of each layer; as output: the stiffness matrix of the laminate constituted of the n layers. Apply this procedure to recover the results of Exercises 14.1 to 14.4.
290
14.6 Implement a numerical procedure having:
as inputs: the stiffness matrix of a given laminate, the in-plane resultants and the resultant moments; as outputs: the inverted stiffness matrix, the in-plane strains and curvatures, the in-plane strains in the material directions of each layer, the in-plane stresses in the material directions of each layer. This procedure will be connected with the procedure implemented in the preceding exercise. Apply these procedures to the case where the laminates of Exercises 14.1 and 14.2 are subjected to the resultants and moments with values: N x = 2.5 kN/mm, N y = 1.5 kN/mm, N xy = 1 kN/mm, M x = 20 Nm/mm, M y = 15 Nm/mm, M xy = 10 Nm/mm.
CHAPTER 15
Effect of the Stacking Sequence. Mat and Cloth Reinforced Materials
When a laminate is constituted of layers which are arranged in an arbitrary stacking sequence, which depends on the nature of layers, orientations, thicknesses of the layers, etc., the stiffness matrix of the laminate has the general form given by Expression (14.29). In many cases, it is desired to arrange the stacking sequence such that a number of terms in the stiffness matrix will be zero, and thus to avoid undesirable coupling between in-plane stretching, bending and twisting. The first part of this chapter is concerned with the analysis of particular cases of laminates for which the stiffness matrix has a simplified form. The analysis will be carried out by following an approximately increasing order of complexity. The types of laminates studied will be a good reference for the usual laminates. Quite often, the elaboration of laminates is done by starting from layers that have the same characteristics (the same constituents, the same thicknesses, etc.), but have different orientations of their material directions with respect to the reference system of the laminate. Particular attention will therefore be given to this type of material. The second part of the chapter will treat the characteristics of cloth reinforced layers, in relation with the parameters of the cloth reinforcement.
15.1 EFFECT OF THE STACKING SEQUENCE 15.1.1 Case of One Layer

15.1.1.1 Isotropic Layer
In the case of a plate constituted of an isotropic homogeneous material, the elastic behaviour is described by the Youngs modulus E and the Poisson ratio. The reduced stiffness matrix for a plane stress state is:
292
Chapter 15 Effect of the Stacking Sequence. Mat and Cloth Reinforced Materials
E 1 2 E Q= 2 1 0
E 1 2
E 1 2 0
0 . E 2 (1 + ) 0
(15.1)
The stiffness matrices of the layer are deduced from Expressions (14.31) to (14.33). Thus:
A11 = A12 = Ee 1 2 = A,
D11 =
12 (1 2 )
Ee3
= D,
Ee = A , 1 2
Bij = 0 ,
D12 = D ,
D22 = D11 = D , D16 = D26 = 0 ,
A22 = A11 = A , A16 = A26 = 0 , Ee 1 A66 = = A, 2 (1 + ) 2
(15.2)
D66
Ee3 1 = = D, 24 (1 + ) 2
with D = A
where e is the plate thickness. It results that the constitutive equation is:
e2 , 12
0 0 0 0 A A 0 xx Nx A A 0 0 0 0 0 N y yy 1 0 A 0 0 0 0 N xy 0 2 xy . = 0 0 0 D D Mx 0 x M 0 0 0 0 D D y y 1 M xy 0 0 0 0 0 D xy 2
0 0 the in-plane strains xx , 0 yy , xy
(15.3)
This equation shows that the in-plane resultants (Nx, Ny, Nxy) depend only on
and the resultant moments (Mx, My, Mxy)
depend only on the curvatures (x, y, xy). In the case of an isotropic plate there is no coupling between in-plane behaviour and flexural behaviour of the plate.
15.1 Effect of the Stacking Sequence
293
15.1.1.2 Orthotropic Layer Referred to its Material Directions

For an orthotropic layer of thickness e, the material directions of which are the same as the reference directions of the plate (the reference directions of the stresses and strains applied to the plate), the reduced stiffness matrix is written:
Q11 Q12 Q = Q12 Q22 0 0
0 0 , Q66
ET
2 1 LT
with
Q11 = EL
2 1 LT
ET EL ET EL
Q22 =
ET EL
ET Q11 , EL
Q12 =
LT ET
2 1 LT
= LT Q22 ,
Q66 = GLT .
Whence the expressions of the stiffness coefficients of the laminate, deduced from Relations (14.31) to (14.33):
A11 = Q11e , A12 = Q12e , A22 = Q22e , A16 = A26 = 0 , A66 = Q66e ,
D11 = Q11
e3 , 12
e3 D12 = Q12 , 12 Bij = 0 , D22 = Q22 e3 , 12 (15.4)
D16 = D26 = 0 ,
D66 with Dij = Aij e2 . 12 0 0 0 D12 D22 0
e3 = Q66 , 12
The constitutive equation of the plate is thus written: N x A11 N y A12 N xy = 0 Mx 0 My 0 M xy 0 A12 A22 0 0 0 0 0 0 A66 0 0 0 0 0 0 D11 D12 0
0 0 xx 0 0 yy 0 0 xy . 0 x 0 y D66 xy
(15.5)
294
As in the case of an anisotropic plate, the in-plane resultants depend only on the in-plane strains and the moments depend only on the curvatures.
15.1.1.3 Orthotropic Layer off Its Material Directions

In the case where the material directions of an orthotropic plate do not coincide with the reference directions of the stresses and strains applied to the plate, the reduced stiffness matrix is: Q12 Q16 Q11 Q22 Q26 , Q = Q12 Q Q Q 26 66 16 are given in Tableau 11.6 as functions of the where the off-axis coefficients Qij coefficients Qij in the material directions. The stiffness coefficients of the plate are expressed as follows: e, Aij = Qij Bij = 0, Dij = Qij e3 e2 = Aij . 12 12
0 0 xx 0 0 yy 0 0 xy . D16 x D26 y D66 xy
(15.6)
The constitutive equation of an orthotropic plate is thus written: N x A11 N y A12 N xy = A16 Mx 0 My 0 0 M xy A12 A22 A26 0 0 0 A16 A26 A66 0 0 0 0 0 0 D11 D12 D16 0 0 0 D12 D22 D26
(15.7)
It is observed again the absence of coupling between in-plane and flexural behaviours. Nevertheless, contrary to the case of an isotropic or an orthotropic plate in its material directions, we observe that the extensional resultants (Nx, Ny)
0 and 0 depend on the extensional strains xx yy , as well as the in-plane shear strain
0 xy . Thus there exists in this case a tension-shear coupling.
In the same way, the components of the moments all depend on the bending curvatures x, y, and on the twisting curvature xy. Thus there also exists a bending-twisting coupling. The effect of the tension-shear coupling can be illustrated by applying a displacement (u, 0, 0) to an orthotropic plate the direction L of which makes an angle with the strain reference directions (Figure 15.1). Figure 15.2 show the deformation of the plate deduced from a finite element analysis, in the case of a unidirectional material of orientation = 45 with the x-direction. The tensionshear coupling introduces a deformed shape in S of the plate.
295
u x
FIGURE 15.1. Plate subjected to tension along the x-direction.
FIGURE 15.2. Deformed shape of a unidirectional plate subjected to tension at 45 to the fibre direction.
15.1.2 Symmetric Laminates

15.1.2.1 General Case
A laminate is symmetric (Chapter 3) if the middle plane is a symmetry plane. Two symmetric layers have:
the same reduced stiffness matrix Qij k , the same thickness ek,
opposite z coordinates zk and zk. It follows from this that the coefficients Bij of the stiffness matrix of the laminate are zero. The constitutive equation is of the general form:
296
N x A11 N y A12 N xy = A16 Mx 0 My 0 M xy 0
A12 A22 A26
A16 A26 A66
0 0 0
D11 D12 D16
0 0 0
D12 D22 D26
0 0 0
0 0 0
0 0 xx 0 0 yy 0 0 xy . D16 x D26 y D66 xy
(15.8)
The stiffness matrix is of the same form as that obtained in (15.7). Thus there does not exist stretching-flexural coupling in the case of symmetric laminates. From this it results that the behaviour of symmetric laminates is simpler to analyze that that of laminates having a stretching-flexural behaviour. In addition, the symmetric laminates do not have the tendency to warp (to bend and twist) that results from the thermal contractions induced during the cooling and curing processes of nonsymmetric laminates. Symmetric laminates are thus largely used, unless some specific conditions require nonsymmetric laminates. For example, a laminate used as a heat shield and exposed to a heat source on only one of its faces will be designed as a nonsymmetric material structure.
15.1.2.2 Symmetric Laminates with Multiple Orthotropic Layers

We consider the case of symmetric laminates constituted of orthotropic layers that have material directions aligned with the laminate directions. The reduced stiffness matrix of each layer has the form:
k k Q11 Q12 k k Q22 Q k = Q12 0 0
0 0 , k Q66
where the reduced stiffness coefficients are expressed as functions of the engineering moduli of each layer according to the relations:
k Q11 = k EL k k 2 ET 1 LT k EL k k ET LT k k 2 ET 1 LT k EL
k Q22 =
k ET k k 2 ET 1 LT k EL
k ET k EL
k Q11 ,
(15.9)
k Q12 =
k k Q22 = LT ,
k k Q66 = GLT .
The stiffness coefficients of the laminate are thus expressed as follows:
297
A11 = A12 = A22 =
k =1 n
k Q11 ek
D11 = D12 = Bij = 0 , D22 =
k =1 n
2 k Q11 ek zk
3 ek + , 12
k =1 n
k Q12 ek ,
3 2 ek k Q12 ek zk + , 12 k =1 2 k Q22 ek zk 3 ek + , (15.10) 12
k =1
k Q22 ek
k =1
A16 = A26 = 0 ,
D16 = D26 = 0 ,
A66 =
k =1
k Q66 ek
D66 =
k =1
k 2 Q66 ek zk
3 ek + . 12
Hence the constitutive equation of the laminate:

N x A11 N y A12 N xy = 0 Mx 0 My 0 M xy 0 A12 A22
0 0
A66
0 0 0
D11 D12
0 0 0
D12 D22
0 0 0 0
0 0 0
0 0 xx 0 0 yy 0 0 xy . 0 x 0 y D66 xy
(15.11)
An equation similar to (15.5) is again obtained. Besides the absence of coupling between in-plane and flexural behaviours, there does not exist tensionshear and bending-twisting couplings.
15.1.3 Antisymmetric laminates

Symmetric laminates are used to avoid coupling between in-plane and flexural behaviours. In contrast, some applications require the use of nonsymmetric laminates. For example, the coupling between in-plane and flexural behaviours may be necessary in the design of turbines with blades that have a warped form. Also, in the case where better shear stiffness has to be obtained, it is necessary to use layers with different orientations. An antisymmetric laminate is constituted of an even number layers of which the distribution of the thicknesses is symmetric, and that of the orientations is antisymmetric with respect to the middle plane. Two layers with symmetric z coordinates thus have: opposed z coordinates zk and zk, the same thickness ek, orientations and with respect to the reference directions of the laminate.
298
The reduced stiffness matrix of the layer of orientation is:

Q +
+ Q11 + = Q12 Q 16+
+ Q12 + Q22 + Q26
+ Q16 + Q26 + Q66
The matrix for the orientation is:

Q11 Q = Q12 Q 16 Q12 Q22 Q26 Q16 Q26 Q66
with
= Q16 + , Q16 = Q26 + , Q26 = Qij + if ij = 11, 12, 22, 66. Qij (15.12)
Whence the stiffness coefficients of an antisymmetric laminate constituted of n = 2p layers: Aij = 2 Aij = 0
Qijk ek
k =1
if ij = 11, 12, 22, 66, if ij = 16, 26,
Bij = 0 Bij = 2
if ij = 11, 12, 22, 66,
Qijk ek zk
k =1 p
if ij = 16, 26, if ij = 11, 12, 22, 66, if ij = 16, 26.
(15.13)
Dij = 2 Dij = 0
3 2 ek k ek zk Qij + 12 k =1
The constitutive equation of an antisymmetric laminate is thus written in the form: N x A11 N y A12 N xy = 0 Mx 0 My 0 M xy B16 A12 A22 0 0 0 B26 0 0 A66 B16 B26 0 0 0 B16 D11 D12 0 0 0 B26 D12 D22 0
0 B16 xx 0 B26 yy 0 0 xy . 0 x 0 y D66 xy
(15.14)
299
y C B A x y C A x u (a) C' D
x
(b)
D D'
FIGURE 15.3. Plate subjected to tension in the x-direction: (a) imposed displacement and (b) imposed stress.
The constitutive equation (15.14) shows that there exists a coupling between in-plane behaviour and twisting, which results from the terms B16 and B26. the effect of this coupling can be illustrated by applying to a plate constituted of an antisymmetric laminate, clamped on the side AB (Figure 15.3), either a displaycement (u, 0, 0) of the side CD (figure 15.3a) or a stress (x, 0, 0) on the face CD (figure 15.3b). Figures 15.4a and 15.4b show the deformed shapes obtained in the two modes of loading. The deformations are amplified so as to show the effects of the coupling. These effects lead to a twisting deformation of the plate which is superimposed on the tensile deformation of the plate.
15.1.4 Cross-Ply Laminates

A cross-ply laminate is constituted of layers the material directions of which are oriented alternately at 0 and 90 with respect to the reference directions of the laminate (Figure 15.5). The reduced stiffness matrix of the 0 layers is: Q11 Q12 Q0 = Q12 Q22 0 0 0 0 . Q66 (15.15)
300
(a)
(b)
FIGURE 15.4. Tension-twisting coupling in the case of an antisymmetric laminate: (a) imposed and (b) imposed stress.
301
Figure 15.5. Cross-ply laminate.
The reduced stiffness constants of the 90 layers are:

= Q22 , Q11 = Q11 , Q22 = Q12 , Q12 = 0, Q26 = 0, Q16 = Q66 , Q66
(15.16)
and the stiffness matrix of the 90 layer is thus:

Q90
Q22 = Q12 0
Q12 Q11 0
0 0 . Q66
(15.17)
The laminate stiffness coefficients Aij, Bij and Dij deduced from Expressions (14.31) to (14.33) for a cross-ply laminate are reported in Table 15.1. Taking into account these results, the constitutive equation thus has the form: N x A11 N y A12 N xy = 0 M x B 11 M y B12 0 M xy A12 A22 0 B12 B22 0 0 0 A66 0 0 B66 B11 B12 0 D11 D12 0 B12 B22 0 D12 D22 0
0 0 xx 0 0 yy 0 B66 xy . 0 x 0 y D66 xy
(15.18)
15.1.4.2 Particular Practical Cases

One particular case, but one of great practical importance, is the case in which the 0 layers have the same thickness, and the 90 layers also have the same thickness, but not necessarily of the same thickness as that of the 0 layers. It is
302
TABLE 15.1. Expressions for stiffness constants of a cross-ply laminate.
A11 = Q11e0 + Q22e90, A22 = Q22 e0 + Q11e90,
A12 = Q12e, A26 = 0,
A16 = 0, A66 = Q66e,
with e0 = cumulative thickness of the 0 layers, e90 = cumulative thickness of the 90 layers, e = thickness of laminate
B11 = Q11b0 + Q22b90, B22 = Q22b0 + Q11b90,
e = e0 +e90 .
B16 = 0, B66 = Q66b,
B12 = Q12b, B26 = 0,
with b0 =
0 layers
ep z p ,
n
b90 =
90 layers
eq zq ,
b=
ek zk = b0 + b90 .
k =1
D11 = Q11d 0 + Q22 d90, D22 = Q22 d 0 + Q11d90,
D12 = Q12 d , D26 = 0,
D16 = 0, D66 = Q66 d ,
with
e3 p + ep z2 d0 = p , 12 0 layers
d90
n
3 eq 2 + eq zq , = 12 90 layers
d=
3 ek 2 + ek zk = d0 + d90 . 12 k =1
always possible to choose the x-direction of the laminate in such a way that it coincides with the 0 direction of the bottom layer of the laminate. From this it results that the 0 layers are the odd layers and the 90 layers are the even layers. If the total number of layers is odd (Figure 15.6a), the laminate is symmetric. If the total number of layers is even (Figure 15.6b), the laminate is said to be antisymmetric. Nevertheless it is not an antisymmetric laminate in the sense of the laminates considered in Subsection 15.1.2. A cross-ply laminate is characterized by the total number n of 0 and 90 layers, and by the ratio of the
303
n even n odd
0 90 0 90 0 5 4 3 2 1 90 0 90 0 90 0 6 5 4 3 2 1
(a) symmetric
(b) antisymmetric
FIGURE 15.6. Symmetric and antisymmetric cross-ply laminates.
cumulative 0 layers to the cumulative thickness of the 90 layers: Re = e0 . e90 (15.19)
In the case where the thicknesses of the layers are the same (the laminates are called regular cross-ply laminates): Re = 1, the stiffness coefficients can then be expressed (Subsections 15.1.4.3 and 15.1.4.4) as functions of n, Re and of the ratio between the moduli: RQ = Q22 ET . = Q11 EL (15.20)
15.1.4.3 Symmetric Cross-Ply Laminates

In the case of symmetric cross-ply laminates (an odd number of layers), the terms Bij are zero, in agreement with the properties of symmetric laminates. The constitutive equation of symmetric cross-ply laminates combines Relations (15.8) and (15.18). So: N x A11 N y A12 N xy = 0 Mx 0 My 0 0 M xy A12 A22 0 0 0 0 0 0 A66 0 0 0 0 0 0 D11 D12 0 0 0 0 D12 D22 0
0 0 xx 0 0 yy 0 0 xy . 0 x 0 y D66 xy
(15.21)
In addition to the coupling between the in-plane and flexural behaviour of symmetric laminates, there also does not exist tension-shear coupling. So, the behaviour of symmetric cross-ply laminates is the same as that of an orthotropic plate in its material directions (15.5). The expressions for the stiffness constants as functions of the number n of layers, of Re and RQ are reported in Table 15.2.
304
TABLE 15.2. Expressions for stiffness constants of a symmetric (odd number of layers) cross-ply laminate.
A11 =
A22 =
1 ( Re + RQ ) Q11e, 1 + Re
A12 = Q12 e,
A16 = 0,
1 + Re RQ 1 1 + Re RQ ) Q11e = A11, ( 1 + Re Re + RQ
A66 = Q66e .
A26 = 0,
Bij = 0,
i, j = 1, 2, 6.
Q11e3 1 + Re A11e 2 D11 = ( RQ 1) + 1 = ( RQ 1) + 1 , 12 Re + RQ 12
D12 =
Q12e3 , 12
D16 = 0,
Q11e3 1 + Re A11e 2 D22 = 1 R + R = 1 R + R , Q) Q Q) Q ( ( 12 Re + RQ 12
D26 = 0, with
D66 =
Q66e3 , 12
1 + Re (n 3) [ Re (n 1) + 2(n + 1) ] (n 2 1) (1 + Re )
3
(1 + Re )
15.1.4.4 Antisymmetric Cross-Ply Laminates

In the case of an antisymmetric cross-ply laminate (an even number of layers), the stiffness constants expressed as functions of n, Re and RQ are reported in Table 15.3. The results of this table show that the constitutive equation then has the form: N x A11 N y A12 N xy = 0 M x B11 My 0 M xy 0 A12 A22 0 0 B11 0 0 0 A66 0 0 0 B11 0 0 D11 D12 0 0 B11 0 D12 D22 0
0 0 xx 0 0 yy 0 0 xy . 0 x 0 y D66 xy
(15.22)
305
TABLE 15.3. Stiffness constants of an antisymmetric (even number of layers) cross-ply laminate.
A11 = A22 =
1 ( Re + RQ ) Q11e, 1 + Re
A12 = Q12e,
A16 = 0,
1 + Re RQ 1 1 + Re RQ ) Q11e = A11, ( 1 + Re Re + RQ Re ( RQ 1) n (1 + Re )
2
A26 = 0,
A66 = Q66e .
B11 =
Q11e 2 =
n (1 + Re ) ( Re + RQ )
Re ( RQ 1)
A11e,
B12 = B16 = 0,
B22 = B11 ,
B26 = B66 = 0.
Q11e3 1 + Re A11e2 D11 = ( RQ 1) + 1 = ( RQ 1) + 1 , 12 Re + RQ 12
D12 =
Q12e3 , 12
D16 = 0,
Q11e3 1 + Re A11e 2 D22 = 1 R + R = 1 R + R , ( ) ( ) Q Q Q Q 12 Re + RQ 12
D26 = 0, with
D66 =
Q66e3 , 12
8Re ( Re 1) 1 + . 1 + Re n 2 (1 + Re )3
Re = 1,
In the case where all the layers have the same thickness:
=1 . 2
In the case where the 0 and 90 layers have the same thickness: Re = 1, and in this case:
A22 = A11 , D22 = D11.
(15.23)
The constitutive equation (15.22) shows that these laminates have only a tensionbending coupling. The effect of this coupling can be illustrated by imposing to a plate constituted of an antisymmetric cross-ply laminate both types of conditions already considered (Figure 15.3). Figures 15.7 show the deformed shapes obtained in both cases. A bending deformation is superimposed on the tensile deformation of the plate. It is also important to note that the coupling coefficient B11 is (Table 15.3) inversely proportional to the total number of layers. From this it results that the tension-bending coupling decreases rapidly when the number of layers is increased.
306
(a)
(b)
FIGURE 15.7. Tension-bending coupling in the case of an antisymmetric cross-ply laminate: (a) imposed displacement and (b) imposed stresses.
15.1.5 Angle-Ply Laminates

An angle-ply laminate is constituted of layers oriented at an angle and layers oriented at an angle . An angle-ply laminate can be arbitrary, symmetric or antisymmetric (Figure 15.8). In fact, the angle-ply laminates of interest are special ones where the layers are oriented alternately in the directions and , with respect to the reference directions of the laminate. Figure 15.8 gives an example of antisymmetric laminate
307
+45 +20 45 20 arbitrary angle-ply
+20 20 20 +20 symmetric angle-ply
+20 20 +20 20 antisymmetric angle-ply
FIGURE 15.8. Different types of angle-ply laminates.
and Figure 15.9 gives an example of symmetric laminate. The constitutive equations of angle-ply laminates are deduced from the general equations (14.29), (15.8) and (15.14).
15.1.5.2 Particular Laminates

In the practical case where the layers have the same thickness, the stiffness constants of the laminates are expressed as functions of the reduced stiffnesses of each layer, of the number n of layers and of the thickness e of the laminate. Qij In the case of layers constituted of the same material, the reduced stiffness constants of the layers are related by the expressions:
= Q11 + , Q11 = Q22 + , Q22 = Q12 + , Q12 = Q26 + , Q26 = Q16 + , Q16 = Q66 + , Q66
(15.24)
+ and the constants Qij referred to where the relations between the constants Qij + are referred to the material directions are given in Table 11.6. The constants Qij to the even layers. For the symmetric the odd layers, and the constants Qij angle-ply laminate of Figure 15.8 and for the symmetric laminate with alternated + = Qij + 20 ; whereas angles of the layers (Figure 15.9), + is equal to 20 and Qij in the case of the antisymmetric laminate of Figure 15.8, + = 20 and + = Qij 20 . Qij
+20 20 +20 20 +20 symmetric angle-ply FIGURE 15.9. Symmetric angle-ply laminate with alternated angles in the layers.
308
1. Symmetric angle-ply laminates
A symmetric angle-ply laminate of the type of Figure 15.8 has an odd number of layers. The stiffness constants are deduced from Relations (14.23), (14.24) and (14.27), associated with Relations (15.24). They are written as: Aij = eQij Aij = 0 Bij = 0 e3 Qij 12 Dij = 0 Dij = if ij = 11, 12, 22, 66, if ij = 16, 26. i, j = 1, 2, 6. if ij = 11, 12, 22, 66, si ij = 16, 26. (15.25)
2. Antisymmetric angle-ply laminates with alternated angles
For an antisymmetric angle-ply laminate with alternated angles (Figure 15.8), the number of layers is even and Relations (15.13) and (15.24) lead to:
Aij = eQij Aij = 0 Bij = 0 Bij = Dij = e2 Qij 2n if ij = 11, 12, 22, 66, if ij = 16, 26. if ij = 11, 12, 22, 66, if ij = 16, 26. if ij = 11, 12, 22, 66, if ij = 16, 26. (15.26)
e3 Qij 12 Dij = 0
3. Symmetric angle-ply laminates with alternated angles
Lastly, in the case of symmetric angle-ply laminate with alternated angles (Figure 15.9), the number of layers is even. The stiffness constants are given by:
Aij = eQij Aij = 0 Bij = 0 Dij = Dij = e3 Qij 12 e3 3n 2 2 Qij 12 n3 if ij = 11, 12, 22, 66, if ij = 16, 26. i, j = 1, 2, 6. if ij = 11, 12, 22, 66, if ij = 16, 26.
(15.27)
We observe that the coefficients Aij are independent of the number of layers of the laminate. The coefficients A16 and A26 are zero.
309
The bending and twisting coefficients D11, D12 and D66 are also independent of the number of layers of the laminate. The coefficients D16 and D26 of bendingtwisting coupling are zero in the case of symmetric angle-ply laminates and of antisymmetric laminates with alternated angles. These coefficients decrease with the number of layers in the other cases. In the case of symmetric angle-ply laminates, there does not exist coupling between the in-plane and flexural behaviours (Bij = 0). In the case of antisymetric angle-ply laminates, this coupling decreases when the number of layers increases.
15.1.6 Laminates with Isotropic Layers

The reduced stiffness of an isotropic layer is given by Relation (15.1). The stiffness constants of a laminate constituted of n isotropic layers with different properties are then expresses by the following relations: A11 =
1 k
k =1 n
Ek ek
, 2
A26 = 0,
A12 = A66 = B12 = B66 =
1kk k2k ,
k =1 n
Ee
E e
A16 = 0,
A22 = A11 ,
2 (1k+k k ).
k =1 n
B11 =
1 k2 ,
k =1
Ek ek zk
k Ek ek zk , 2 k =1 1 k
E e z
B16 = 0,
B22 = B11 ,
B26 = 0,
k k k . 2 (1 + k ) k =1 n
(15.28)
3 Ek 2 ek , D11 = e z + 2 k k 12 k =1 1 k
3 k Ek 2 ek , D12 = e z + 2 k k 12 k =1 1 k
D22 = D11 ,
D16 = D26 = 0,
D66
3 2 ek Ek , = ek zk + 2 (1 + k ) 12 k =1
The constitutive equation of a laminate with isotropic layers is thus written as: N x A11 N y A12 N xy = 0 M x B11 M y B12 M 0 xy A12 A22 0 B12 B22 0 0 0 A66 0 0 B66 B11 B12 0 D11 D12 0 B12 B22 0 D12 D22 0
0 0 xx 0 0 yy 0 B66 xy . 0 x 0 y D66 xy
(15.29)
310
The constitutive equation has the same form as the constitutive equation (15.18) of a cross-ply laminate. In the case of a symmetric laminate with isotropic layers, the constitutive equation simplifies to the general equation (15.8) of symmetric laminates, and is written: 0 0 0 0 0 xx N x A11 A12 0 N 0 0 0 0 A A y 12 22 yy 0 N 0 0 0 0 xy A66 xy = 0 (15.30) , Mx 0 0 0 D11 D12 0 x My 0 0 0 D12 D22 0 y M 0 0 0 0 0 D xy 66 xy the stiffness coefficients Aij and Dij being given by Relations (15.28).
15.1.7 Arbitrary Laminates

The various laminates considered in the previous subsections show the influence of the arrangement of layers one upon another and with respect to the reference directions of the laminates. The particular cases studied correspond to important applications. The laminates considered have shown the various coupling which are induced between tension, shear, bending and twisting. In the case of laminates with arbitrary stacking, the constitutive equation of the laminate is written in the general form (14.29). In this case, the different couplings are induced simultaneously. The effect of these couplings can be illustrated by imposing upon a plate made of an arbitrary laminate a tensile loading (Figure 15.3) with an imposed displacement or imposed stresses. Figures 15.10a and 15.10b show the results obtained. Shear, bending and twisting deformations are superimposed on the tensile deformation. These couplings are generally undesirable.
15.2 ANALYSIS OF MAT AND CLOTH REINFORCED MATERIALS 15.2.1 Introduction

The fundamental two-dimensional orthogonal weaves used in the fabrication of laminates and sandwiches have been introduced in Chapter 2. The principal parameters for the characterization of a weave fabric are: the type of weave style le type (unidirectional, taffeta, satin, twill, etc.), the nature of the warp and weft fibres: glass, carbon, Kevlar, stratifil, roving, etc.,
311
(a)
(b)
FIGURE 15.10. General coupling in the case of an arbitrary laminate: (a) imposed displacement and (b) imposed stresses.
312
the linear density (weight per unit length) of the fibres expressed in tex, the number of the warp and weft fibres, respectively, per unit length and width of the cloth, the weight per unit area, etc. The experimental values obtained show that the elastic behaviour of a reinforced cloth composite is little dependent on the type of weave style, contrary to the damage characteristics. The purpose of Section 15.2 is to establish an evaluation of the moduli of reinforced two-dimensional cloth composite based on a laminate analogy.
15.2.2 Caracterization of a Cloth Reinforcement

A two-dimensional orthogonal weave fabric is formed (Chapter 2) by threads woven in two directions (Figure 15.11): the warp direction (corresponding to the material L of the orthotropic composite layer) and the warp direction (corresponding to the material direction T of the layer). The warp is characterized by: nwp : the number of threads per unit width (Figure 15.11), Twp : the linear density (weight per unit length of the threads). The weight of the warp fibres per unit width is therefore: mwp = nwpTwp , and the volume of the warp fibres per unit width may be expressed as: (15.31)
v wp =
mwp
wp
nwpTwp
wp
(15.32)
where wp is the fibre density of the warp fibres. Similarly, the weft is characterized by: nwf : the number of threads per unit length (Figure 15.11), Twf : the linear density (weight per unit width of the threads). The weight of the warp fibres per unit length is:
mwf = nwf Twf ,
(15.33)
and the volume of the warp fibres per unit length is :

v wf =
mwf
wf
nwf Twf
wf
(15.34)
on introducing the fibre density wf in the weft direction.
15.2 Analysis of Mat and Cloth Reinforced Materials
313
weft warp L T
nwp
nwf
1m 1m
FIGURE 15.11. Schematization of a two-dimensional cloth.
The relative proportion of the fibres in the warp direction, called as the balancing coefficient in the warp direction, is given by: nwpTwp v wp wp = . (15.35) k= v wp + v wf nwpTwp nwf Twf +
wp
wf
In the case of fibres of the same nature in both warp and weft directions ( wp = wf ), the balancing coefficient is written as: k= nwpTwp nwpTwp + nwf Twf . (15.36)
In the case of identical linear densities in the warp and weft directions, the expression for k reduces to: nwp k= . (15.37) nwp + nwf In the case of fibres of the same nature, we distinguish different types of cloths according to the value of the balancing coefficient k: if k = 1, the cloth is unidirectional in the warp direction, if k = 0, the cloth is unidirectional in the weft direction, if k = 1/2, the cloth is balanced in the warp and weft directions.
314
The cloth reinforcement is also characterized by its weight per unit area Ms. At the time of manufacture of a laminate, the cloth is impregnated by matrix to form a layer of thickness ec. The volume fraction of fibres can then be calculated by considering the volume vt of a layer of unit area:
vt = ec. The volume of fibres contained in the volume vt is: vf =
Ms
(15.38)
on introducing the density of the fibres. From this it results that the volume fraction of the fibres is expressed as: v M 1 Vf = f = s . (15.39) vt f ec By expressing the volume fraction in the preceding relation as a function of the weight fraction Pf of the fibres (1.19), the layer thickness is given by:
ec = M s Pf m + (1 Pf ) f , Pf f m
(15.40)
where m is the density of the matrix.
15.2.3 Laminate Analogy

The laminate analogy, used to model the elastic behaviour of a cloth reinforced layer, consists in considering the layer as constituted of two unidirectional layers (Figure 15.12): one warp layer oriented at 0 and one weft layer oriented at 90. The respective thicknesses of these layers are considered to be proportional to the volume fractions of the fibres: the thickness hwp of the warp layer is proportional to Vwp and the thickness hwf of the weft layer is proportional to Vwf.
T L hwp 0 90 weft warp hwf
FIGURE 15.12. Laminate analogy of a cloth layer.
315
These thicknesses are expressed as function of the thickness ec of the layer and of the balancing coefficient in the warp direction, as follows: hwp = kec , hwf = (1 k ) ec . The elastic behaviour of each layer can be described by their moduli: warp layer: weft layer: ELwp ELwf ETwp ETwf (15.41) (15.42) GLTwp , GLTwf .
LTwp LTwf
These moduli are referred to the respective material directions of each layer: the L direction in the warp direction for the warp layer, the L direction in the weft direction for the weft layer. From this, it results that the reduced stiffness matrix of the warp layer, referred to the material directions of this layer, and thus to those of the cloth layer, is expressed from (11.52) as follows:
wp Q11 = wp ELwp , wp Q16 = 0, wp Q22 = wp ETwp , wp Q12 = wp LTwp ETwp , wp Q26 = 0, ch Q66 = GLTwp ,
(15.43)
with
wp =
1
2 1 LT wp
ETwp ELwp
(15.44)
Similarly, the stiffness constants of the weft layer are expressed in its material directions as:
wf Q11 = wf ELwf , wf Q16 = 0, wf Q22 = wf ETwf , wf Q12 = wf LTwf ETwf , wf Q26 = 0, wf Q66 = GLTwf ,
(15.45)
with
wf =
1
2 1 LT wf
ETwf ELwf
(15.46)
The reduced stiffness constants of the weft layer, referred to the material directions of the cloth layer, are thus:
wf wf wf = Q22 wf = Q12 wf = 0, Q11 , Q12 , Q16 wf wf = Q11 Q22 , wf wf = Q66 Q66 .
wf = 0, Q26
(15.47)
From Expression (14.31), the stiffness constants Aij that describe the in-plane
316
behaviour of the cloth layer are written:

wp tr Aij = hwpQij + hwf Qwfij ,
(15.48)
or
wp wf . Aij = ec kQij + (1 k ) Qij
(15.49)
Thus, by considering Expressions (15.43) to (15.47), we obtain: A11 = ec k wp ELwp + (1 k ) wf ETwf , A12 = ec k wp LTwp ETwp + (1 k ) wf LTwf ETwf , A22 = ec k wp ETwp + (1 k ) wf ELwf , A66 = ec kGLTwp + (1 k ) GLTwf , A16 = A16 = 0. These relations neglect the curvature (Figure 15.11) of the threads that results from the weaving. Halpin, Jerine and Whitney [19] have studied the influence of this curvature, which leads to a reduction in these constants. Similarly, the preceding analysis does not take into account a possible misalignment on the threads, that also leads to a decrease in the properties. (15.50)
15.2.4 In-Plane Behaviour of a Cloth Reinforced Layer

The in-plane behaviour of a cloth reinforced layer is described by the constitutive equation: N x A11 N y = A12 N xy 0 A12 A11 0
0 0 xx 0 0 yy , 0 A66 xy
(15.51)
where the coefficients Aij are given in (15.50).

1. Tension in the warp direction (L direction)
In the case of a tension in the warp direction, the in-plane resultants are given as follows: N x 0, Thus: N y = 0, N xy = 0. (15.52)
317
0 N x = A11 xx + A12 0 yy , 0 0 = A12 xx + A22 0 yy , 0 0 = xy .
(15.53)
Whence the expression of the in-plane resultant: A2 0 N x = A11 12 xx . A 22 The Youngs modulus EL in the warp direction may be written as:
EL =
(15.54)
xx N x ec = 0 . 0 xx xx
2 1 A12 A . 11 ec A 22
(15.55)
It results that the modulus is expressed as: EL = (15.56)
From Expressions (15.53), the strain in the weft direction is given by the relation: A12 0 0 xx . (15.57) yy = A22 It results that the Poisson ratio LT is expressed by the relation:
LT =
A12 . A22
(15.58)
2. Tension in the weft direction (T direction)
The previous results can be transposed to the case of a tension in the weft direction. It results that the Youngs modulus ET in the weft direction is given by:
2 1 A12 ET = A22 , ec A11
(15.59)
and the Poisson ratio TL is expressed as:
TL =
3. In-Plane shear
A12 E = LT T . A11 EL
(15.60)
A shear test in the material directions of the cloth reinforced layer leads to the
318
determination of the in-plane shear modulus GLT of the layer. Thus: GLT = GTL = 1 A66 . ec (15.61)
15.2.5 In-Plane Moduli of a Cloth Reinforced Layer

Expressions (15.56) to (15.61) can be regrouped as follows:
2 1 A12 EL = A11 , ec A22
ET =
1 A2 A22 12 , ec A11 A12 , A22
(15.62) A12 E = LT T , A11 EL
LT =
TL =
1 A66 . ec
GLT = GTL =
The Expressions of the Youngs moduli can be simplified as: 1 A11 , ec 1 ET = (1 ) A22 , ec EL = (1 ) by introducing the parameter
2 A12 , A11 A22
(15.63) (15.64)
(15.65)
the value of which is practically equal to zero. On taking Expressions (15.50) for the coefficients Aij into account, the moduli can be written in the form:
EL = (1 ) k wp ELwp + (1 k ) wf ETwf , ET = (1 ) k wp ETwp + (1 k ) wf ELwf , k wp LTwp ETwp + (1 k ) wf LTwf ELwf k wp ETwp + (1 k ) wf ELwf (15.66) ,
LT =
GLT = kGLTwp + (1 k ) GLTwf ,
319
with k wp LTwp ETwp + (1 k ) wf LTwf ELwf . = k wp ELwp + (1 k ) wf ETwf k wp ETwp + (1 k ) wf ELwf Expressions (15.66) give the in-plane moduli of a two-dimensional cloth reinforced layer. The moduli ELwp, ETwp, . . . , ELwf, ETwf, . . . can themselves be expressed (Chapter 9) as functions of the characteristics of the fibres (Ef, f) and the matrix (Em, m), of the balancing coefficient k, of the weight Ms of the cloth and the thickness ec of the layer. Expressions (15.66) can be simplified in the case where the fibres in the warp and weft directions are identical, and in the case where the cloth is balanced. In fact, in this case, k = 1/2, and the moduli in the warp and weft directions are identical: ELwp = ELwf = ELu , ETwp = ETwf = ETu ,
2
LTwp = LTwf = LTu ,

GLTwp = GLTwf = GLTu ,
(15.67)
where ELu, ETu, LTu and GLTu are the moduli of a unidirectional layer having a fibre volume fraction equal to that of the cloth reinforced layer under consideration. Expressions (15.66) for the moduli then reduce to:
EL = ET = 1 (1 ) u ( ELu + ETu ) , 2
LT =
2 LTu , E Lu 1+ ETu
(15.68)
GLT = GLTu ,
with
2 4 LT u
E Lu 1 + E Tu
u =
1
2 1 LT u
ETu E Lu
(15.69)
The expressions for and u show that:
EL = ET
1 ( ELu + ETu ) . 2
(15.70)
In the case of an unbalanced cloth of threads of the same nature in the warp and weft directions, the values of the moduli must be calculated from the general expressions (15.66). The exploitation of these expressions is complex to put into practice. An approximation can be given for them by introducing in (15.66) the moduli (15.67). The moduli of the cloth layer are then expressed as follows:
320
EL = (1 ) u kELu + (1 k ) ETu , ET = (1 ) u kETu + (1 k ) ELu , 1 k + (1 k ) E Lu ETu
LT =
LTu ,
(15.71)
GLT = GLTu , with
2 LT u
E Lu E Lu k E + 1 k k + (1 k ) E Tu Tu
u =
1
2 1 LT u
ETu E Lu
(15.72)
15.2.6 Numerical Applications

We consider the case of two cloths: a balanced cloth and an unbalanced cloth, made of glass fibres. The properties of the constituents are: glass fibres: matrix:
Ef = 73 GPa, Em = 3 GPa,
f = 0.22, m = 0.35,
f = 2,600 kg/m3 , m = 1,200 kg/m3 .
1. Layer with a balanced cloth
The cloth reinforcement considered has a weight per unit area equal 500 g/m2 and the thickness of the layer is 0.7 mm. The volume fraction of the fibres determined by (15.39) leads to: Vf = 27.5 %, corresponding to a weight fraction Pf = 45.1 %. Using the results developed in Chapter 9, we obtain:
ELu = 22.2 GPa, GLTu = 1.87 GPa,
LTu = 0.31,
ETu = 5.12 GPa.
(15.73)
Whence the characteristics of the balanced cloth reinforced layer are:

EL = ET = 13.8 GPa,
LT = 0.12,
GLT = 1.87 GPa .
(15.74)
2. Layer with an unbalanced cloth
The cloth reinforcement has a weight per unit area of 650 g/m2, and the unbalance in the warp direction is characterized by k = 0.6. A layer is made of the same weight fraction as the preceding layer made of the balanced cloth. The thickness of the layer made can be deduced from (15.39). We obtain: ec = 0.91 mm.
321
The characteristics of the layer are obtained from the same values (15.73) for the unidirectional composite with a volume fraction of 27.5 %. Relations (15.72) lead to the values:
EL = 15.5 GPa, ET = 12.1 GPa, GLT = 1.87 GPa.
LT = 0.135,
(15.75)
15.2.7 Mat Reinforced Layer

A mat (Chapter 2) is made of chopped fibres or continuous fibres randomly oriented in the plane of the mat. A mat reinforced layer can then be considered as a laminate made of an infinite number of layers oriented in all the directions. A layer of orientation containing fibres oriented between and + d, and thus, by the analogy introduced in Subsection 15.2.3, has a thickness equal to: ec (15.76) d . 2
In the case of a mat reinforcement, Relation (14.31) for the in-plane coefficients Aij is therefore expressed as: Aij = ec 2
=0
) d . ( Qij
(15.77)
For evaluating these integrals, it is interesting to use the expressions in Table 11.7, since the integrals of sin 2, sin 4, cos 2, cos 4 are zero. We obtain: A11 = ecV1 , A22 = ecV1 , A12 = ecV4 , A26 = 0, A16 = 0, (15.78)
e A66 = ecV5 = c (V1 V4 ) . 2 with

V1 = 1 3Q + 3Q22 + 2Q12 + 4Q66 ) , 8 ( 11 V4 = 1 Q + Q22 + 6Q12 4Q66 ) , 8 ( 11 V5 =
1 2 1 Q + Q 2Q + 4Q (V1 V4 ) = 8 ( 11 22 12 66 ) ,
(15.79)
Q11 = u ELu , Q12 = u LTu ETu , 1

2 1 LT u
Q22 = u ETu , Q66 = GLTu , ETu E Lu , (15.80)
u =
322
where ELu, ETu, LTu and GLTu are the moduli, introduced in (15.67), of a unidirectional layer with fibre volume fraction equal to that of the mat reinforced layer considered. The in-pane moduli of the mat reinforced layer are then deduced from Relations (15.62). Thus:
ELmat = ETmat =
(V1 V4 )(V1 + V4 ) ,
V1
LTmat =
V4 , V1
1 2
(15.81)
GLTmat = V5 =
(V1 V4 ) .
ELmat . 2 (1 + LTmat )
We verify that:
GLTmat =
(15.82)
From this relation, it results that the Youngs modulus Ex, given by Relation (11.9), does not depend upon the direction in which the modulus is measured. The mat layer behaves in the plane of the layer as an isotropic material. As a numerical application, we consider the case of a layer reinforced with a mat of weight per unit area equal to 450 g/m2 and the thickness of which is 1 mm. The characteristics of the materials are the same as those considered in Subsection 15.2.6. The fibre volume fraction calculated from (15.39) is: Vf = 17.3 % This leads to the moduli of the unidirectional material being equal to:
ELu = 15.1 GPa, GLTu = 1.54 GPa,
LTu = 0.327,
ETu = 4.30 GPa.
(15.83)
The characteristics of the mat reinforced layer determined from (15.81) are thus:
ELmat = ETmat = 7.72 GPa, GLTmat = 2.91 GPa.
LTmat = 0.33,
(15.84)
15.2.8 Laminate with Cloth and Mat Reinforced Layers

The results established in Section 15.1 may be applied to laminates made of cloth and mat reinforced layers. In this subsection we treat as examples the case of two symmetric laminates (Figure 15.13) made of four cloth layer and four mat layers. The cloth reinforced layer has the characteristics determined in (15.74):
323
cloth cloth mat mat mat mat cloth cloth
1.4 mm 2 mm cloth cloth cloth cloth
mat mat
2 mm 1.4 mm
mat mat
(a)
(b)
FIGURE 15.13. Laminates made of cloth and mat reinforced layers.
EL = ET = 13.8 GPa, ec = 0.7 mm.
LT = 0.12,
GLT = 1.87 GPa,
(15.85)
And the mat reinforced layer has the characteristics determined in Relations (15.84): ELmat = ETmat = 7.72 GPa, LTmat = 0.33, GLTmat = 2.91 GPa, (15.86) ec = 1 mm. The reduced stiffness constants of the layers are deduced from Relations (11.52). Whence: for the cloth reinforced layers: Q11 = Q22 = EL
2 1 LT
= 14.0 GPa, (15.87)
Q12 = LT Q22 = 1.68 GPa, Q66 = GLT = 1.87 GPa, Q16 = Q26 = 0. for the mat reinforced layers:
m m = Q22 = Q11
ELmat
2 1 LT mat
= 8.66 GPa, (15.88)
m m = LTmat Q22 = 2.86 GPa, Q12 m = GLTmat = 2.91 GPa, Q66 m m = Q26 = 0. Q16
324
As the laminates are symmetric, the stiffness matrices of the constitutive equation of the laminates reduce to the in-plane stiffness matrix and the flexural matrix. The coupling matrix is zero. Moreover, in the calculations, the adjacent layers of the same nature can be regrouped into a single layer.
1. Laminate with external cloth reinforced layers (Figure 15.13a)
From Relation (14.31), the in-plane stiffness coefficients Aij may be written in the form:
m Aij = 2 Qij 2 103 + Qij 1.4 103 .
(15.89)
Whence
A11 = 73.858 106 N/m, A22 = A11 , A16 = 0, A12 = 16.140 106 N/m, A66 = 16.876 106 N/m, A26 = 0.
(15.90)
From Relation (14.33), the flexural stiffness coefficients Dij are written as:
3 m 23 9 2 1.4 Dij = 2 Qij 2 + 12 + Qij 1.4 2.7 + 12 10 .
(15.91)
Whence
D11 = 338.41 Nm, D22 = D11, D16 = 0,
D12 = 50.312 Nm, D66 = 54.546 Nm, D26 = 0.

0 xx 0 0 0 0 yy 0 0 0 0 xy 338.41 50.312 0 x 50.312 338.41 0 y xy 0 0 54.546
(15.92)
The constitutive equation of the laminate is therefore:

N x 73.858 106 16.140 106 0 6 6 0 N y 16.140 10 73.858 10 N 0 0 16.876 106 xy = Mx 0 0 0 My 0 0 0 0 0 0 M xy 0 0 0
(15.93)
2. Laminate with external mat reinforced layers (Figure 15.1a)
The difference in the stacking sequence modifies only the flexural stiffnesses Dij, which here are expressed, from (14.33), as:
3 3 9 2 1.4 m 2 2 + + + 1.4 0.7 2 2.4 Dij = 2 Qij Q 10 . ij 12 12
(15,94)
Exercises
325
Thus: D11 = 236.77 Nm, D22 = D11, D16 = 0, D12 = 72.755 Nm, D66 = 73.347 Nm, D26 = 0. (15.95)
The constitutive equation of the laminate is thus written as:

N x 73.858 106 16.140 106 0 6 6 0 N y 16.140 10 73.858 10 N 0 0 16.876 106 xy = Mx 0 0 0 My 0 0 0 0 0 0 M xy
0 xx 0 0 0 0 yy 0 0 0 0 xy 236.77 72.755 0 x 72.755 236.77 0 y 0 0 74.347 xy
(15.96) Comparison of Expressions (15.93) and (15.96) shows that the in-plane behaviours of the two laminates are identical (the terms Aij are equal). The terms D11 are D22 are higher in the case where the cloth reinforced layers are external, that leads to greater bending stiffnesses. This property results from the layers with higher moduli (EL and ET) which are the most distant from the middle plane. In contrast, the bending-twisting coupling stiffness D12 and the twisting stiffness D66 are higher in the case where the mat reinforced layers are external.
EXERCICES
15.1 Establish the expressions in Table 15.2. 15.2 Establish the expressions in Table 15.3. 15.3 We consider the [0/90]kS symmetric cross-ply laminates of a given thickness h, made of 4k layers oriented at 0 and 90. The layers have the same thickness h/4k and the same mechanical characteristics. How does the stiffness matrix change when k increases from k = 1 (laminate [0/90]S)? 15.4 We consider the [0/90]k antisymmetric cross-ply laminates of a given thickness h, made of 2k layers oriented alternately at 0 and 90. The layers have the same thickness h/4k and the same mechanical characteristics. How does the stiffness matrix change when k increases from 1 (laminate [0/90])?
326
15.5 Do the previous exercise again for the case of [90/0]k antisymmetric laminates where the layers are oriented alternately at 90 and 0. Compare with the preceding results. 15.6 A hybrid cloth is woven from carbon rovings in the warp direction and glass rovings in the weft direction. The warp roving has 600 filaments per unit width of the cloth. The filaments have a weight per unit length of 1,000 tex and a density of 1,800 kg/m3. The mechanical characteristics of the filaments are Ef = 250 GPa and f = 0.32. The weft roving has 200 filaments per unit length. The filaments have a weight of 2,400 tex and a density of 2,600 kg/m3. The mechanical characteristics of the filaments are Ef = 72 GPa and f = 0.22. Calculate the weight per unit area of the cloth and the balancing coefficient. The cloth is used with a matrix of mechanical characteristics Em = 3.5 GPa and m = 0.30 to obtain a layer of thickness 1.4 mm. Calculate the volume fractions of the fibres in the warp direction and in the weft direction. Using the laminate analogy, calculate the warp and weft moduli, and then the in-plane stiffnesses and the in-plane moduli of the cloth reinforced layer.
CHAPTER 16
Governing Equations and Energy Formulation of the Classical Laminate Theory
16.1 GOVERNING EQUATIONS 16.1.1 General Relations

The governing equations of the classical laminate theory are obtained by introducing the constitutive equation (14.29) of laminates into the plate equations (13.52) or (13.57) in the case of dynamics problems, or into equations (13.59) in the case of statics problems. Substituting, for example, Equation (14.29) into Relation (13.57), then by taking account of Expressions (14.15) relating strains to displacements, we obtain the three governing equations of the classical laminate theory:
A11 2 u0 2u0 2u0 2v0 2v0 2v0 + 2 A16 + A66 + A16 2 + ( A12 + A66 ) + A26 2 xy xy x 2 y 2 x y 3w 0 x3
2 u0 t 2
B11
3B16
3w 0 x 2y
( B12 + 2 B66 )
3w 0 xy 2
B26
3w 0 y 3
(16.1)
= s
,
2u0 2u0 2v0 2v0 2v0 + A26 + + + A 2 A A 66 26 22 xy xy y 2 x 2 y 2 3w 0 x 2 y 3B26 3w 0 xy 2 B22 3w 0 y 3
A16
2u0 x 2
+ ( A12 + A66 )
B16
3w 0 x3
2v0 t 2
( B12 + 2 B66 )
(16.2)
= s
328
Chapitre 16 Governing Equations and Energy Formulation of Laminate Theory
D11
4w 0 x 4
+ 4 D16
4w 0 x3y
+ 2 ( D12 + 2 D66 )
4w 0 x 2y 2 3u0 xy 2
+ 4 D26
4w 0 xy 3
+ D22
4w 0 y 4
B11
3u0 x3
3B16
3u0 x 2y
( B12 + 2 B66 ) 3v0 xy 2
B26
3u0 y 3
B16
3v0 x3
( B12 + 2 B66 ) = q s 2w 0 t 2
3v0 x 2y
3B26
B22
3v0 y 3
(16.3)
The preceding equations do not take into account the body forces, the possible shear stresses on the faces of the laminate, and neglect the effects of rotation inertia. The accounting for these factors leads us to introduce additional terms in Equations (16.1) to (16.3) in accordance to Relations (13.52). Equations (16.1) to (16.3) constitute the governing relations of the classical laminate theory. These equations, associated with the boundary conditions imposed at the structure (Section 16.2), allow us to find, in principle, the in-plane displacements u0(x, y, t), v0(x, y, t) and w0(x, y, t), which are the solutions of the elasticity problem. Solving these equations is, however, complex, and can be done analytically only in some particular cases. An important simplification appears when the coupling terms Bij are zero (the case of symmetric laminates, for example). In this case, Equations (16.1) to (16.3) are partially decoupled: Equations (16.1) and (16.2) contain only the displaycements u0(x, y, t) and v0(x, y, t), whereas Equation (16.3) introduces only the displacement w 0(x, y, t). In all other cases, it is necessary to solve coupled equations.
16.1.2 Symmetric Laminate

In the case where the laminate is symmetric, all the coupling terms Bij are zero (Chapter 15), as well as the quantities R. The governing equations are then written in the form:
A11 2 u0 2u0 2u0 2v0 2v0 2v0 + 2 A16 + A66 + A16 2 + ( A12 + A66 ) + A26 2 xy xy x 2 y 2 x y 2 u0 t 2
= s
(16.4)
16.1 Governing Equations
329
A16
2u0 x 2 = s
+ ( A12 + A66 )
2u0 2u0 2v0 2v0 2v0 2 A A A + A26 + + + 66 26 22 xy xy y 2 x 2 y 2
2v0 t 2
,
4w 0 x3y + 2 ( D12 + 2 D66 ) 4w 0 x 2y 2 + 4 D26 4w 0 xy 3 + D22
(16.5)
4w 0 y 4
D11
4w 0 x 4
+ 4 D16
= q s
2w 0 t 2
4w 0 4w 0 + I xy 2 2 + 2 2 . x t y t
(16.6)
In the case of statics problems, the relations reduce to:

A11 2 u0 x 2 + 2 A16 2u0 2 u0 2v0 + A66 + A 16 xy y 2 x 2 v0 v + A26 20 = 0, xy y
2 2
(16.7)
+ ( A12 + A66 )
A16
2u0 x 2
+ ( A12 + A66 )
2u0 2 u0 2v0 + A26 + A 66 xy y 2 x 2 2v0 2v + 2 A26 + A22 20 = 0, xy y
(16.8)
D11
4w 0 x 4
+ 4 D16
4w 0 x3y
+ 2 ( D12 + 2 D66 ) 4w 0 xy 3
4w 0 x 2y 2 4w 0 y 4 = q.
(16.9)
+ 4 D26
+ D22
The in-plane behaviour (u0, v0) is decoupled from the bending behaviour (w0). If, in addition, the laminate is a balanced laminate: A16 = A26 = 0, Relations (16.4), (16.5) or (16.7), (16.8) are simplified, Relation (16.6) or (16.9) remaining unchanged.
16.1.3 Antisymmetric Cross-Ply Laminates

As another example, we consider the case of an antisymmetric cross-ply laminate, for which (Subsection 15.1.4.4) the in-plane stiffnesses are A11, A12, A22 = A11, A66, the flexural stiffnesses are D11, D12, D22 = D11, D66, and the coupling stiffnesses B11, B22 = B11. In the case of a statics problem, the governing equations are written:
330
A11
2 u0 x 2
+ A66
2u0 y 2
+ ( A12 + A66 )
2v0 3w 0 B11 = 0, xy y 3
(16.10) (16.11)
( A12 + A66 )
2u0 2v 2v 3w 0 + A66 20 + A11 20 + B11 = 0, xy x y y 3
4w 3u0 3v0 4w 0 4w 0 D11 40 + + 2 D + 2 D B 3 = q. (16.12) ( ) 12 66 11 4 2 2 3 x y y x y x Due to the introduction of the terms B11 and B22 = B11, the differential equations are coupled.
16.1.4 Expressions for Resultants and Moments

The expressions for the resultants and moments as functions of the displaycements are obtained by substituting Expressions (14.15) for the strains into the constitutive equation of the laminate. We obtain: u u v v 2w 0 N x = A11 0 + A16 0 + 0 + A12 0 B11 x x y x 2 y (16.13) 2w 0 2w 0 2 B16 B12 , xy y 2
N y = A12
u u0 v v 2w 0 + A26 0 + 0 + A22 0 B12 x x y x 2 y
2w 0 2w 0 , 2 B26 B22 xy y 2
N xy = A16 u u0 v v 2w 0 + A66 0 + 0 + A26 0 B16 x x y x 2 y
(16.14)
2w 0 2w 0 2 B66 B26 , xy y 2 M x = B11 u u0 v v 2w 0 + B16 0 + 0 + B12 0 D11 x x y x 2 y
(16.15)
2w 0 2w 0 2 D16 D12 , xy y 2 M y = B12 u u0 v v 2w 0 + B26 0 + 0 + B22 0 D12 x x y x 2 y
(16.16)
2w 0 2w 0 2 D26 D22 , xy y 2
(16.17)
16.1 Governing Equations
331
M xy = B16
u u0 v v 2w 0 + B66 0 + 0 + B26 0 D16 x x y x 2 y
2w 0 2w 0 2 D66 D26 . xy y 2
(16.18)
The expressions for the transverse shear resultants are obtained by substituting the preceding expressions for the moments into Equations (13.56):
Qx = B11 2 u0 2 u0 2 u0 2v0 + 2 B16 + B66 + B16 2 xy x 2 y 2 x 2v0 2v 3w 0 3w 0 + B26 20 D11 3 D 16 xy y x3 x 2y 3w 0 xy 2 D26 3w 0 y 3 ,
+ ( B12 + B66 )
(16.19)
( D12 + 2 D16 ) 2u0 x 2
Q y = B16
+ ( B12 + B66 )
2 u0 2 u0 2v0 B + B26 + 66 xy y 2 x 2
+ 2 B26 3D26
2v0 2v 3w 0 3w 0 2 + B22 20 D16 D + D ( ) 12 66 xy y x3 x 2y 3w 0 xy 2 D22 3w 0 y 3 .
(16.20)
The classical laminate theory is based on the assumption of the nullity (14.1) of the transverse shear strains xz and yz. This hypothesis is satisfied only if the transverse shear resultants Qx and Qy are zero. In fact, the preceding relations (16.19) and (16.20) lead to resultants which are not zero. This apparent inconsistency is admitted in the framework of the classical laminate of plates.
16.1.5 Expressions for Stresses

The expressions for the in-plane stresses as functions of the displacement are obtained by substituting Relations (14.15) for the strains and curvatures into Expressions (14.20) for the stresses. We obtain:
k k = Q11 xx
u0 v k u0 k v0 + Q16 + 0 + Q12 x x y y
2 2 k 2w 0 k w0 k w0 z Q11 + + Q Q 2 , 16 12 2 2 x y x y
(16.21)
332
k k = Q12 yy
u0 v k u0 k v 0 + Q26 + 0 + Q22 x x y y
2 2 k 2w 0 k w0 k w0 z Q12 + + Q Q 2 , 26 22 2 2 x y x y
(16.22)
k k = Q16 xy
u0 v k u0 k v 0 + Q66 + 0 + Q26 x x y y
2 2 k 2w 0 k w0 k w0 z Q16 + + Q Q 2 . 66 26 2 2 x y x y
(16.23)
The transverse shear stresses xz and yz can be derived from the fundamental equations (13.20) of the mechanics of materials. By substituting the preceding expressions (16.21) to (16.23) for the in-plane stresses, and then integrating them over the variable z through the thickness of each layer. We obtain in the case of statics problems:
3 3 3w 0 z 2 k 3w 0 k w0 k k k w0 + + + + 3 2 Q Q Q Q Q11 16 12 66 26 2 x3 x 2y xy 2 y 3
k = xz
2 2 2 k 2 u0 k u0 k u0 k v0 Q Q Q 2 z Q11 + + + 16 66 16 xy x 2 y 2 x 2 k k + Q12 + 2Q66
(16.24)
v0 k v0 k + Q26 + a ( x, y ), ) 2 xy y
k = yz
3 3 3w 0 z 2 k 3w 0 k k k w0 k w0 + + + + 2 3 Q Q Q Q Q16 12 66 26 22 2 x3 x 2y xy 2 y 3
2 2 2 k 2u0 k k u0 k u0 k v0 Q Q Q Q 2 z Q16 + + + + 12 66 26 66 xy x 2 y 2 x 2
(16.25)
k + 2Q26
2 2v0 k v0 + Q22 + b k ( x, y ). 2 xy y
Functions ak(x, y) and bk(x, y) are functions of integration, determined by expressing the continuity of the transverse shear stresses xz and yz between the layers and making zero the transverse shear stresses on the lower and upper faces of the laminate.
16.2 Boundary Conditions
333
16.2 BOUNDARY CONDITIONS 16.2.1 Basics

The conditions imposed at the boundary of a structure are those that guarantee unique solutions of the governing equations (16.1) to (16.3). As these equations are fourth-order partial differential equations in x and y, four conditions have to be imposed at the boundary of the structure. A boundary element (Figure 16.1) is described at point P(x, y, 0) of the boundary by its unit normal n and the tangential unit vector t in the middle plane. The deformed shape of the laminate at point P is characterized by the displacement of point P expressed in the basis (n , t , k ) by its components: u0 n ( x, y ), u0t ( x, y ), w 0 ( x, y ) and by the orientw 0 . The loads exerted at point P tation of the deformed shape characterized by n are characterized by the in-plane resultants Nn, Nnt, the transverse shear resultant Qn, and the moments of bending Mn and twisting Mnt. The conditions prescribed at point P take on one of the quantities of each of the following pairs:
u0 n , N n ; The quantity
u0t , Nt ;
w 0 , Mn; n
w0 ,
M nt + Qn . t
(16.26)
M nt + Qn is known to be the Kirchhoffs boudary condition. t The prescribed values (generally zero) will be overlined in the following subsections.
16.2.2 Simply Supported Edge

In the case of a simply supported edge (Figure 16.2), usually four possibilities are retained and are classified as follows: z
M nt
t
Mn n
x
Figure 16.1. Plate edge element.
334
t
P
n
Figure 16.2. Simply supported edge.
A1 : A2 : A3 : A4 :
w 0 = 0, w 0 = 0, w 0 = 0, w 0 = 0,
M n = 0, M n = 0, M n = 0, M n = 0,
u0n = u0n , Nn = Nn , u0n = u0 n , Nn = Nn ,
u0t = u0t , u0t = u0t , N nt = N nt , N nt = N nt . (16.27)
In practice, the condition usually retained is:

w 0 = 0,
M n = 0, N n = 0, N nt = 0.
(16.28)
16.2.3 Clamped Edge

In the case of a clamped edge (Figure 16.3), four possibilities are also considered: E1 : E2 : E3 : E4 : w 0 = 0, w 0 = 0, w 0 = 0, w 0 = 0,
w 0 = 0, n w 0 = 0, n w 0 = 0, n w 0 = 0, n w 0 = 0, n
u0n = u0n , Nn = Nn , u0 n = u0n , Nn = Nn ,
u0t = u0t , u0t = u0t , (16.29) N nt = N nt , N nt = N nt .
In practice, the condition retained is: w 0 = 0, u0n = 0, u0t = 0. (16.30)
16.2 Boundary Conditions
335
t
P
n
Figure 16.3. Clamped edge.
16.2.4 Free Edge

In the case of a free element (Figure 16.4), the exerted load is zero, and then the resultants and moments Nn, Nnt, Qn, Mn and Mnt are zero. It results that five conditions have to be satisfied, although only four conditions are necessary. To remove this difficulty, it is usual to proceed in the following way. The twisting couple which is exerted on a boundary element of length dt is considered as the composition of two forces of resultants Mnt and Mnt + dMnt the supports of which are distant of dt. The variation of the moment Mnt is given by:
M nt dt . t The equilibrium of the boundary element under the effect of the twisting couple and the transverse shear resultant Qn dt is written as:
d M nt =
M nt + M nt +
M nt d t + Qn d t = 0 , t
t
M nt + d M nt M nt
dt
Figure 16.4. Free element.
336
or
M nt + Qn = 0 . t
(16.31)
This relation constitutes the Kirchhoffs boundary condition. And so the boundary conditions at a free element are written as:
N n = 0,
N nt = 0,
M n = 0,
M nt + Qn = 0. t
(16.32)
16.3 ENERGY FORMULATION OF LAMINATE THEORY 16.3.1 Introduction

The energy theorems (Section 8.3) can be used to derive a variational formulation of the governing equations of laminates. This formulation, associated with the boundary conditions, provides (Section 8.4) the bases for the development of approximate solutions of the mechanical behaviour of laminates. The energy theorems are also the bases for the analysis of composite structures by finite element method.
16.3.2 Strain Energy of a Laminate

The strain energy (8.50) of an elastic solid is written in Cartesian coordinates as follows:
Ud = 1 2
xx xx
+ yy yy + zz zz + yz yz + xz xz + xy xy ) d x d y d z ,
(16.33) where the integration is extended over the entire volume of the solid. Taking into account the assumptions of the laminate theory, zz = 0, xz = yz = 0, and Relations (14.19) expressing the stresses as functions of the strains, the strain energy may be written as:
Ud =
1 2
(Q
k 2 11 xx
k 2 k 2 k + Q22 yy + Q66 xy + 2Q12 xx yy
k xx xy 2Q16
k + 2Q26 yy xy
) d x d y d z.
(16.34)
This relation can be expressed as a function of the displacements u0, v0 and w0,
16.3 Energy Formulation of Laminate Theory
337
by substituting into the preceding expression the strain-displacement relations (14.14) and (14.15), established in Chapter 14. Next, by integrating with respect to z through the thickness of the laminate, we obtain:
1 Ud = 2
2 u 2 u0 v0 v0 0 + A22 A11 + 2 A12 x x y y 2
u v u v v u + 2 A16 0 + A26 0 0 + 0 + A66 0 + 0 x y y x x y B11 B22 v0 2w 0 u0 2w 0 u0 2w 0 2 B 12 y x 2 + x y 2 x x 2
2w 0 u0 v0 v0 2w 0 u0 2w 0 + + 2 B 2 1 6 2 y y 2 x x xy x y
2w 0 u0 v0 v0 2w 0 2B26 2 + +2 x y xy y y 2w 0 2w 0 u0 v0 2w 0 2w 0 4B66 + + + D D 2 11 12 y 2 xy x x 2 y 2 x 2w 0 2w 0 2w 0 2w 0 + D22 + + 4 D D 26 y 2 16 x 2 y 2 xy w0 + 4 D66 xy

2 2 2 2
(16.35)
d x d y.
In this expression, the integration has to be implemented in the middle plane of the laminated plate. This expression introduces the stiffnesses Aij, Bij and Dij respectively given by Relations (14.23), (14.24) and (14.27). The expression for the strain energy introduces an in-plane strain energy associated with the in-plane stiffnesses Aij and a flexural strain energy associated with the flexural stiffnesses Dij. The expression of the strain energy contains also coupling terms between the in-plane displacements u0, v0 and the transverse displacement w0. As in the case of the governing equations (Section 16.1), these coupling terms are introduced by the coupling stiffnesses Bij. In the case of symmetric laminates, the coupling terms Bij between in-plane behaviour and flexural behaviour are zero, and Expression (16.35) of the strain energy is reduced to the following:
338
1 Ud = 2
2 u 2 u0 v0 v0 0 A11 + A22 + 2 A12 x x y y 2
u v u v v u + 2 A16 0 + A26 0 0 + 0 + A66 0 + 0 x y y x x y 1 + 2
dx dy
2 2 2 2 2w 0 2w 0 w w 0 0 D11 x 2 + 2 D12 x 2 y 2 + D22 y 2 2
(16.36)
2w 0 2w 0 2w 0 2w 0 + 4 + + D D 4 D 26 66 16 x 2 xy y 2 xy
d x d y.
The strain energy appears as the sum of two terms: the first term contains only the in-plane displacement u0 and v0, the second term contains only the transverse displacement w0. In the case of problems of pure flexural behaviour of laminate, the first term reduces to a constant C, and the strain energy may be written as follows: 1 Ud = 2
2 2 2 2 2w 0 2w 0 w w 0 0 D11 x 2 + 2 D12 x 2 y 2 + D22 y 2 2 2 2 2 2
(16.37)
w0 w0 w0 w0 D D D + 4 + + 4 16 26 66 2 2 xy xy d x d y + C. x y
Moreover, if the laminate is such that D16 = D26 = 0 (the case of orthotropic laminate with material confounded with the directions x and y of the plate), the strain energy further simplifies to: 1 Ud = 2
2 2 2 2 2w 0 2w 0 w w 0 0 D11 x 2 + 2 D12 x 2 y 2 + D22 y 2 2 2
(16.38)
w0 + 4 D66 xy d x d y + C.
16.3.3 Kinetic Energy of a Laminate

The kinetic energy of a structure (8.54) is written as:
1 Ec = 2 u 2 v 2 w 2 + + dx dy dz , t t t
(16.39)
16.3 Energy Formulation of Laminate Theory
339
where is the density of the material at a point (x, y, z). The integration is implemented over the whole volume of the structure. In the case of the classical laminate theory, the displacement field (14.3) is written:
w 0 , x w v = v0 z 0 , x w = w 0 ( x, y ) , u = u0 z
(16.40)
where w0 is independent of z. On substituting these relations into Expression (16.39), the kinetic energy of the laminate may be written as:
1 Ec = 2
2 2 2 2 v0 w 0 2 u w w 0 0 0 z + z + dx dy dz . t t xt y t t
(16.41) Neglecting the derivatives with respect to time of the laminate rotations (thus neglecting the rotary inertia terms), then integrating through the laminate thickness, the expression of the kinetic energy of the laminate is reduced to:
Ec = 1 2
u0 2 v0 2 w 0 2 + + dx dy , t t t
(16.42)
introducing the weight per unit area (13.29) of the laminate at point (x, y).
16.3.4 Work of External Loads

In the case of transverse bending, the loads can be reduced to the transverse loads exerted on the lower and upper faces of the laminate. The variation of the work of these loads is written as:
Wf =
zz
( h 2 ) zz ( h 2 ) w 0 d x d y .
(16.43)
Therefore by introducing the stresses q defined in (13.38):

Wf =
q w
q w
0
dx dy .
(16.44)
The energy function Wf of the external loads is then expressed as:

Wf = dx dy .
(16.45)
CHAPTER 17
Including Transverse Shear Deformation in Laminate Theory
17.1 LIMITATION OF THE LAMINATE THEORY

The classical laminate theory (Chapter 14), based on Kirchhoffs hypothesis, allows us to describe, with a good precision, the stress and strain fields in laminates that are not very thick, except in the little extended region near the free edges of the laminates. The validity of the laminate theory has been established by comparing the results deduced from this theory with experimental results and with the exact solutions of the governing equations, solutions which can be derived in some particular configurations (Chapter 19). In contrast, in the case of thick laminates (a width-to-thickness ratio less than about 10), the results derived from the classical laminate theory show significant differences with the actual mechanical behaviour of the laminates: the deflection of the laminates, the stress distributions, etc. A first improvement consists of the introduction of the effect of the transverse shear deformation, with a first-order theory (Sections 17.2 and 17.3). A second improvement consists in modifying this theory by introducing correction factors for the transverse shear moduli of the laminate. This approach, considered in Section 17.4, is an extension to the case of laminates of theories developed by Reissner [20] and Mindlin [21] in the case of isotropic homogeneous plates. This analysis was first extended to the case of laminates by Whitney and Pagano [22-24].
17.2 STRAIN AND STRESS FIELDS 17.2.1 Field of Displacements

The model used is a first-order theory of the displacement field having the general form (13.3), thus: u ( x, y, z , t ) = u0 ( x, y, t ) + z x ( x, y, t ), v ( x, y, z, t ) = v0 ( x, y, t ) + z y ( x, y, t ), (17.1) w ( x, y, z , t ) = w 0 ( x, y, t ).
17.2 Strain and Stress Fields
341
A H B middle plane
A H B
FIGURE 17.1. Deformation in the case of a first-order model including the effect of the transverse shear.
with
u0 ( x, y, t ) = u ( x, y, 0, t ), v0 ( x, y, t ) = v ( x, y, 0, t ), w 0 ( x, y, t ) = w ( x, y, 0, t ). We have shown (Section 13.2.3) that, in a general first-order model, the line AB normal to the middle plane of the laminate remains straight under deformation. However, contrary to the classical theory, the line AB does not stay normal to the deformed middle plane (Figure 17.1) in the case where the transverse shear effect is taken into account.
17.2.2 Strain Field

The strain field is deduced from the displacement field (17.1). It is written (13.7):
xx = yy = zz =
u u0 = +z x, x x x y v v0 , = +z y y y w w 0 = = 0, z z v w w = + = y + 0 , z y y u w w + = x + 0 , z x x u v u0 v0 x y + = + + . + z y x y x y x
(17.2)
yz = 2 yz
xz = 2 xz = xy = 2 xy =
342
Chapter 17 Including Transverse Shear Deformation in the Laminate Theory
The strain tensor at point M of the laminate is thus: xx ( M ) = xy xz
xy
xz
(17.3)
yy yz , yz 0
and the strain matrix consists of five nonzero components:

xx yy ( M ) = xy . yz xz The strain field can be subdivided into two fields: the in-plane and flexural strain field: xx mf ( M ) = yy , xy the transverse shear strain field:
w 0 +y yz y . s (M ) = = xz w 0 + x x
(17.4)
(17.5)
(17.6)
In practice, the transverse shear field varies from one layer to another through the thickness of the laminate. The preceding expression shows that the first-order theory including the transverse shear effect leads to a transverse shear strain which is independent of the z coordinate. The first approach consists in taking the transverse shear strain field as equal to the mean transverse shear strain field 0 0 yz and xz of the laminate. Thus:
0 yz yz s (M ) = = . 0 xz xz
(17.7)
As in the case of the classical laminate theory, the in-plane and flexural strain field is considered as the superposition:
of the in-plane strains (or midplane strains):
343
u0 0 x xx 0 v 0 m ( M ) = yy = , y 0 xy u0 v0 x + y
(17.8)
which is expressed exclusively as a function of the in-plane displacements (u0, v0) in the middle plane (Oxy) of the points of this plane; and of the flexural (bending and twisting) strains:
z x f x xx f y f ( M ) = yy = z y f xy x y z + x y .
(17.9)
The bending and twisting strains may be expressed as functions of the curvature matrix by the relation: f ( M ) = z ( x, y ) , (17.10) with x x x y ( x, y ) = y = (17.11) . y xy y x+ x y Finally, the field mf(M) is written:
0 xx x = 0 yy + z y , 0 xy xy
mf
with
0 xx =
(17.12)
u0 , x
0 yy = y =
v0 , y y y
,
0 xy =
u0 v0 , + y x
x = x , x
xy
y . = x+ y x
(17.13)
344
The strain field is thus written in a form analogous to Expression (14.16) of the classical laminate theory:
mf ( M ) = m ( x, y ) + z ( x, y ) .
(17.14)
Only the expressions for the curvatures are modified.
17.2.3 Stress Field

The stresses in the layer k are expressed by the general expression (13.12), thus:
Q12 xx Q11 Q Q 22 yy 12 Q26 xy = Q16 0 yz 0 0 xz k 0
Q16 Q26 Q66

0 0
0 0 0 C44 C45
0 0 0 C45 C55 k
xx yy xy . 0 yz 0 xz
(17.15)
and Cij of the layer k are referred to the reference directions of The stiffnesses Qij the laminate. They are expressed as functions of the stiffnesses in the material directions of the layers (Tables 11.3 and 11.6). Their expressions are reported in Table 17.1. The stress field is the superposition of the in-plane stresses: xx, yy, xy, and of the transverse shear stresses: yz, xz. Expression (17.15) shows that these two fields are uncoupled.
TABLE 17.1. Reduced stiffness constants referred to the reference directions of the laminate as functions of the constants referred to the material directions of the layers.
= Q11 cos 4 + Q22 sin 4 + 2 ( Q12 + 2Q66 ) sin 2 cos 2 , Q11
= ( Q11 + Q22 4Q66 ) sin 2 cos 2 + Q12 sin 4 + cos 4 , Q12 = ( Q11 Q12 2Q66 ) sin cos3 + ( Q12 Q22 + 2Q66 ) sin 3 cos , Q16 = Q11 sin 4 + Q22 cos 4 + 2 ( Q12 + 2Q66 ) sin 2 cos 2 , Q22 = ( Q11 Q12 2Q66 ) sin 3 cos + ( Q12 Q22 + 2Q66 ) sin cos3 , Q26
2 2 4 4 = Q66 Q11 + Q22 2 ( Q12 + Q66 ) sin cos + Q66 sin + cos ,
= C44 cos 2 + C55 sin 2 , C44 = ( C55 C44 ) sin cos , C45 = C44 sin 2 + C55 cos 2 . C55
17.3 Fundamental Equations of Laminates Including Transverse Shear Deformation
345
The in-plane stresses in the layer k may be written as:
xx Q11 Q12 Q22 yy = Q12 Q Q26 xy 16 k

or in the condensed form:
Q16 Q26 Q66 k
0 xx Q11 Q12 0 Q22 yy + z Q12 0 Q Q26 xy 16
Q16 Q26 Q66 k
x y , xy
k ( M ) = Q k m ( x , y ) + z Q k ( x, y ) .
(17.16)
These expressions are similar to Relations (14.20) and (14.21) of the classical laminate theory, but differ by Expressions of the curvatures (17.13). The transverse shear stresses in the layer k are expressed as:
yz C44 = xz k C45 C45 C55 k 0 yz . 0 xz
(17.17)
17.3 FUNDAMENTAL EQUATIONS OF LAMINATES INCLUDING TRANSVERSE SHEAR DEFORMATION 17.3.1 Constitutive Equation
The constitutive equation of laminates including the effect of the transverse shear deformation is, by (17.16) and (17.17), the superposition of Relation (14.29) of the classical laminate theory and the equation that involves the transverse shear resultants Qx and Qy introduced in (13.18). Thus:
Qx Q ( x, y ) = = Q y
h
k =1
hk
xz dz . yz k 1 k
(17.18)
According to Expressions (17.17) and (17.18), the equation for the transverse shear resultants is written as:
Qy F44 = Qx F45 F45 0 yz 0 , F55 xz
(17.19)
with
Fij =
( hk hk 1 ) (Cij )k ,
k =1
i, j = 4, 5.
(17.20)
The coefficients Fij have the same form as the coefficients Aij (14.23). They differ
346
instead of the reduced by using the transverse shear stiffness constants Cij (14.23). stiffness constants Qij
The constitutive equation of laminates with transverse shear is written by associating the resultants and the moments in the form:
N x A11 N A y 12 N xy A16 M x = B11 M y B12 M xy B16 Qy 0 Qx 0 with

0 = xx
A12 A22 A26 B12 B22 B26 0 0
A16 A26 A66 B16 B26 B66 0 0
B11 B12 B16 D11 D12 D16 0 0
B12 B22 B26 D12 D22 D26 0 0
B16 B26 B66 D16 D26 D66 0 0
0 0 0 0 0 0 F44 F45
0 0 xx 0 0 yy 0 0 xy 0 x (17.21) 0 y 0 xy F45 0 yz F55 0 xz
u0 , x
0 yy =
x = 0 yz
x , x w = 0 +y , y
v0 , y y y = , y w 0 xz = 0 + x. x
0 = xy
xy
u0 v0 + , y x y = x+ . y x
(17.22)
The constitutive equation can also be written in the condensed form as: N A B 0 m M = B D 0 . f 0 0 Q F c The stiffness constants of the matrices are expressed as: (17.23)
Aij =
) = ( hk hk 1 ) ( Qij
k
1 2 2 ) = Bij = hk hk 1 ( Qij k 2 k =1 Dij = Fij = 1 3 3 ) hk hk 1 ( Qij k 3 k =1
k =1 n
( Qij )k ek ,
k =1
(
n
( Qij )k ek zk ,
k =1 n
3 2 ek Q e z = + ( ij )k k k 12 , k =1
(17.24)
k =1
) = ( hk hk 1 ) ( Cij
k
( Cij )k ek .
k =1
347
TABLE 17.2. Stiffness constants in the material directions as functions of the engineering constants.
Q11 =
EL
2 1 LT
ET EL
Q22 =
ET Q11 , EL
Q22 = LT Q22 , C44 = GTT ,
Q66 = GLT , C55 = GLT .
and Cij , referred to the reference directions of the laminate, The stiffnesses Qij are expressed in Table 17.1, as functions of the stiffnesses Qij and Cij in the material directions of the layers. These stiffness constants are themselves expressed (Chapters 9 and 10) as functions of the six engineering constants: EL, ET, LT, GLT, GLT and GTT . These expressions are remembered in Table 17.2.
The constitutive equation (17.21) and (17.23) is the superposition: of the constitutive equation in the same form (14.30) as the classical laminate theory : N A B m (17.25) M = B D , f (Equations (14.30) and (17.25) differ by the expressions for the curvature matrix in the classical theory and the theory with shear deformation), and of the transverse shear equation:
[Q ] = [ F ][ c ] .
17.3.2 Governing Equations
(17.26)
The governing equations which take account of the effect of the transverse shear deformation are derived by introducing the constitutive equation (17.21) in the fundamental equations (13.50) of plates. We obtain:
A11 2 u0 x 2 + 2 A16 2u0 2u0 2v0 2v0 2v0 + A66 + + + + A A A A ( 12 66 ) 16 26 xy xy y 2 x 2 y 2
+ B11 + B26
2 x x 2 2 y y 2
2 y 2 y 2 x 2 x + 2 B16 + B66 + B16 + ( B12 + B66 ) xy xy y 2 x 2 = s 2u0 t 2 +R 2 x t 2
(17.27)
348
A16
2u0 x 2
+ ( A12 + A66 )
2 u0 2 u0 2v0 2v0 2v0 + A26 + + + A 2 A A 66 26 22 xy xy y 2 x 2 y 2
+ B16
2 x x 2
+ ( B12 + B66 )
y y y 2 x 2 x + B26 + + + B 2 B B 66 26 22 xy xy y 2 x 2 y 2
= s
2v0 t 2
+R
2 y t 2
(17.28)
x 2w 0 x y y 2w 0 2w 0 + + + + + F55 F 2 F 2 x 45 y 44 y + y 2 +q x x y x
= s 2 u0 x 2 2w 0 t 2
,
2u0 2u0 2v0 2v0 2v0 + B66 + + + + B B B B ( ) 16 12 66 26 xy xy y 2 x 2 y 2
(17.29)
B11
+ 2 B16
+ D11 + D26
2 x x 2 2 y y
2
2 y 2 y 2 x 2 x + 2 D16 + D66 + D16 + ( D12 + D66 ) xy xy y 2 x 2 w w F55 x + 0 F45 y + 0 x y

+ I xy 2 x t 2
=R 2u0
2u0 t 2
(17.30)
B16
2 u0 2 u0 2v0 2v0 2v0 + ( B12 + B66 ) + B26 + B66 2 + 2 B26 + B22 2 xy xy x 2 y 2 x y
+ D16 + D22
2 x x 2 2 y y
2
y y 2 x 2 x + ( D12 + D66 ) + D26 + + D 2 D 66 26 xy xy y 2 x 2 w w F45 x + 0 F44 y + 0 x y

+ I xy 2 y t 2
=R
2v0 t 2
(17.31)
Equations (17.27) to (17.31) allow us, in principle, to determine the five functions that are solutions: u0(x, y, t), v0(x, y, t), w0(x, y, t), x(x, y, t) and y(x, y, t). These functions have to satisfy the boundary conditions imposed on the structure studied. In the case of symmetric laminates, Bij = 0 and R = 0, Equations (17.27) and (17.28) respectively reduce to Equations (16.4) and (16.5) of the classical theory.
349
Equation (17.29) is unchanged, whereas Equations (17.30) and (17.31) simplify as follows:
D11 2 x x 2 2 y 2 y 2 x 2 x + 2 D16 + D66 + D16 + ( D12 + D66 ) xy xy y 2 x 2
2
+ D26
= I xy
2 y y
2 x t 2
w w F55 x + 0 F45 y + 0 x y
,
2 2
(17.32)
D16
2 x x 2
y y 2 x 2 x + ( D12 + D66 ) + D26 + + D 2 D 66 26 xy xy y 2 x 2

2
+ D22 = I xy
2 y y 2 y t 2
w w F45 x + 0 F44 y + 0 x y
(17.33)
17.3.3 Boundary Conditions

The boundary conditions are obtained by applying variational theorems to the strain energy of the plates, taking into account the transverse shear deformation. The equations obtained lead to conditions imposed upon one of the variables of each of the following pairs:
u0 n , N n ; u0t , Nt ;
n , M n ;
t , M t ;
w 0 , Qn ,
where n and t are the normal and tangential directions (Figure 16.1) at a point of the boundary where the conditions are imposed. In the case of the theory with transverse shear deformation, five conditions are therefore imposed. We give hereafter some examples.
1. Free edge. For any point P of a free edge (Figure 16.4), the boundary conditions may be written as: N n = 0, Nt = 0, M n = 0, t = 0, w 0 = 0. (17.34) 2. Simply supported edge (Figure 16.2). The conditions are written:
N n = 0, u0t = 0, Nt = 0, M n = 0, M n = 0,
t = 0,
t = 0,
w 0 = 0,
w 0 = 0.
(17.35) (17.36)
or
N n = 0,
350
3. Clamped edge (Figure 16.3).
The boundary conditions may be written as:

u0 n = 0, u0t = 0,
n = 0,
t = 0,
w 0 = 0.
(17.37)
17.3.4 Stresses in Layers

As the functions u0, v0, w0, x and y are determined, the in-plane stresses in layer k are derived from Expression (17.16) as:
u0 x v0 y u0 v0 y + x x x y . y x y + x y
Q12 xx Q11 Q22 yy = Q12 Q26 Q16 xy k
Q16 Q26 k Q66
(17.38)
Q12 Q11 Q22 + z Q12 Q26 Q16
Q16 Q26 k Q66
The transverse shear stresses in layer k are deduced from Expression (17.17). We obtain:
y + 0 C C yz 44 y 45 . = C w 0 55 k xz k C45 x + x w
(17.39)
These expressions show that the transverse shear stresses are uniform in each layer and discontinuous between layers. Such a distribution is not very realistic. A better estimate of the transverse shear stresses can be obtained by substituting Expressions (17.38) for the in-plane stresses in the fundamental equations (8.20) of the mechanics of materials. For example, in the case of a statics problem, Relations (8.20) in layer k are written: k k k xx + xy + xz =0, x y z (17.40)
351
k k k yy + yz + xy = 0. y z x Whence: k k k xz = xx xy , z x y k k k yz = yy xy . z y x By combining Relations (17.38), (17.42) and (17.43), we obtain:
(17.41)
(17.42) (17.43)
k k k xz = a1 ( x, y ) + za2 ( x, y ) , z k k yz = b1k ( x, y ) + zb2 ( x, y ) , z on introducing the functions:

k k ( x, y ) = Q11 a1
(17.44) (17.45)
2u0 x 2
k 2Q16
2 2u0 k u0 Q66 xy y 2
k Q16
2v0 x
2
k k Q12 + Q66
2 2v0 k v0 Q26 , xy y 2
(17.46)
k k a2 ( x, y ) = Q11
2 x x 2
k 2Q16
2 2 x k x Q66 xy y 2 k ) xy Q26
k Q16
2 y x
2
k k Q12 + Q66
2 y
2 y y 2
(17.47) ,
k k ( x, y ) = Q16 b1
2u0 x 2
k k Q12 + Q66
2 2u0 k u0 Q26 xy y 2
k Q66
2v0
2 2 k v0 k v0 Q Q 2 , 26 22 xy x 2 y 2 k k Q12 + Q66
(17.48)
k k b2 ( x, y ) = Q16
2 x x 2
x k x Q26 ) xy y 2
k Q22
k Q66
2 y x
2
k 2Q26
2 y xy
2 y y 2
(17.49)
of layer k are In the expressions of these functions, the reduced stiffnesses Qij
k . The integration of Relations (17.44) and (17.45) next leads to the denoted Qij
352
expressions for the transverse shear stresses in layer k:

k k k xz ( x, y ) = a0 ( x, y ) + za1 ( x, y ) +
k yz ( x,
z2 k a2 ( x, y ), 2
(17.50)
(17.51)
y)
k = b0 ( x,
y) +
k zb1 ( x,
z2 k y ) + b2 ( x, y ). 2
k k The functions of integration a0 ( x, y ) and b0 ( x, y ) are determined by considering:
k k from one layer to the next, the continuity of xz and yz k k and yz on the two outer faces of the laminate. the vanishing of xz
This process for evaluating the transverse shear stresses leads to a parabolic variation of the stresses through the thickness of the laminate.
17.4 MODIFIED THEORY OF LAMINATES WITH TRANSVERSE SHEAR 17.4.1 Introduction of Transverse Shear Coefficients
An improvement of the laminate theory including the transverse shear deformation, developed in Sections 17.2 and 17.3, consists in considering the following model: 1. The part of the constitutive equation (17.21) relating to the in-plane resultants and moments (Nx, Ny, Nxy, Mx, My, Mxy) is not modified. 2. The part relating to the transverse shear resultants is modified by replacing the stiffness coefficients Fij by new transverse shear coefficients Hij of the laminate. Thus the transverse shear resultants are written as:
Qy H 44 Q = H x 45 H 45 0 yz 0 , H 55 xz i, j = 4, 5.
(17.52)
with
H ij = kij Fij ,
(17.53)
The parameters kij are shear correction factors to be determined. The inverse relation expressing the mean strains as functions of the transverse shear resultants are:
0 yz K 44 0 = xz K 45 K 45 Q y . K55 Qx
(17.54)
17.4 Modified Theory of Laminates with Transverse Shear
353
The matrices H ij and Kij are the inverses of each other. For example:
H 44 = H 55 K55 , K K = 44 , K H 45 = H 54 = K = K 45 , K
(17.55)
2 . K 44 K55 K 45
The constitutive equation is written in a form analogous to (17.21) by changing the transverse shear stiffnesses, thus: N x A11 N A y 12 N xy A16 M x = B11 M y B12 M xy B16 Q 0 y 0 Qx A12 A22 A26 B12 B22 B26 0 0 A16 A26 A66 B16 B26 B66 0 0 B11 B12 B16 D11 D12 D16 0 0 B12 B22 B26 D12 D22 D26 0 0 B16 B26 B66 D16 D26 D66 0 0 0 0 0 0 0 0 H 44 H 45 0 0 xx 0 0 yy 0 0 xy 0 x . (17.56) 0 y 0 xy H 45 0 yz H 55 0 xz
The expressions for strains and curvatures are given by Relations (17.22). It results that the governing equations are identical to Equations (17.27) to (17.33), replacing the terms Fij by the new transverse shear stiffnesses (17.55).
17.4.2 Evaluation of the Shear Correction Factors in the Case of Orthotropic Plates
In this subsection, we consider the case of a homogeneous plate made of an orthotropic material the directions of which are the same as the reference directions of the plate. In the case of a pure bending, Equation (17.38) for the in-plane stresses reduces as:
xx Q11 Q12 yy = z Q12 Q22 0 0 xy Therefore, from Relation (17.42) it becomes:
0 x 0 y . Q66 xy
(17.57)
xz = z Q11 x + Q12 y ) + ( Q66 xy ) . ( z x x
(17.58)
354
Integration, taking into account xz ( h 2 ) = 0, leads to:
xz
h2 z 2 = ( Q11 x + Q12 y ) + ( Q66 xy ) 8 2 , y x
(17.59)
where h is the thickness of the plate. The constitutive equation (17.56), considering Expressions (15.4), leads to: h3 M x = ( Q11 x + Q12 y ) , 12 h3 M xy = Q66 xy . 12 Expression (17.59) for the transverse shear stress is thus written as: 3 M x M xy + 2h x y z2 1 4 . h2 (17.60) (17.61)
xz =
(17.62)
Thus, on taking account of Equation (13.56), we obtain:
xz
A similar argument leads to:
3 z2 = Qx . 1 4 2 2h h 3 z2 1 4 2 Qy . 2h h
(17.63)
yz =
(17.64)
A first method to determine the parameters kij consists in expressing the transverse shear resultants using Relation (17.52). In the case of an orthotropic homogeneous plate, this relation is written as: Q y = k44 F44 0 yz ,
0 Qx = k55 F55 xz ,
(17.65) (17.66) (17.67)
with
F44 = hC44 , F55 = hC55 .
The stress-strain relation (10.1) in the case of an orthotropic material is written:
yz = C44 yz ,
xz = C55 xz .
It results that the mean stresses determined by Relation (17.52) are:
Qy k55 h Qx . k44 h
(17.68) (17.69)
0 yz =
0 xz =
(17.70)
355
The coefficients k55 and k44 are determined so as to make these mean stresses equal to the stresses given by (17.63) and (17.64), for z = 0:
yz =
3Qy 2h
xz =
3Qx . 2h
(17.71)
Comparison between Expressions (17.70) and (17.71) leads to:
k44 = k55 = k =
2 . 3
(17.72)
In the same way, we could have adjusted the mean stresses (17.70) to the mean values of the stresses (17.63) and (17.64):
yz = xz
Whence:
1 h
3 z2 1 4 Qy d z , h2 h 2 2h
h2
(17.73)
1 = h
3 z2 1 4 h 2 Qx d z . h 2 2h
h2
(17.74)
yz =
Qy h
xz =
Qx . h
(17.75)
In this case, the adjustment leads to:

k44 = k55 = k = 1 .
(17.76)
We are then in the case of the initial theory (Sections 17.2 and 17.3). A second method consists in considering the strain energy per unit area of the laminate. The strain energy induced by the transverse shear is per unit volume: udc = 1 ( xz xz + yz yz ) . 2
h2 h 2
(17.77)
The strain energy per unit area is thus written as:

U dc =
1 2
( xz xz + yz yz ) d z .
(17.78)
The energy calculated by introducing the mean strains (17.65) and (17.66) is expressed as:
0 U dc =
1 2h
Qy Qx + yz xz dz . k55C55 k44C44 h 2
h2
(17.79)
Thus, on introducing Expressions (17.73) and (17.74) for the stresses, we have:
0 U dc 2 2 Qy 1 Qx = + . k C k C 2h 55 55 44 44
(17.80)
356
In the case of an orthotropic homogeneous plate, the energy can also be calculated by substituting the elasticity relations (17.68) and (17.69) into (17.78): U dc = 1 2
2 2 xz yz + dz . C C h 2 55 44 h2
(17.81)
On introducing the stresses given by (17.63) and (17.64), and then integrating, we obtain: 2 2 Qy 1 6 Qx U dc = + (17.82) . C C 2h 5 55 44 Comparing Expressions (17.80) and (17.82) then leads to: k44 = k55 = k = 5 . 6 (17.83)
17.4.3 Evaluation of the Shear Correction Factors in the Case of a Laminated Plate
17.4.3.1 Shear Stresses
k k k The in-plane stresses xx , yy and xy in layer k are deduced from (17.16) as:
( ) k k 0 k 0 k 0 k k k yy = Q12 xx + Q22 yy + Q26 xy + z ( Q12 x + Q22 y + Q26 xy ) , k k 0 k 0 k 0 k k k xy = Q16 xx + Q26 yy + Q66 xy + z ( Q16 x + Q26 y + Q66 xy ) .
k k 0 k 0 k 0 k k k xx = Q11 xx + Q12 yy + Q16 xy + z Q11 x + Q12 y + Q16 xy ,
(17.84) (17.85) (17.86)
These expressions can be put in the form:

k xx = [Q1 ]k [ m ( x, y ) + z ( x, y ) ] ,
k yy = [Q 2 ]k [ m ( x, y ) + z ( x, y ) ] , k xy = [Q6 ]k [ m ( x, y ) + z ( x, y )] ,
(17.87) (17.88) (17.89)
where m is the in-plane strain matrix (17.8), is the curvature matrix (17.11), and the matrices [Qi ]k defined in each layer are introduced as follows:
Qk Qk Qk , [Q1 ]k = 11 12 16 Qk Qk Qk , [Q 2 ]k = 12 22 26
(17.90) (17.91)
357
Qk Qk Qk . [Q6 ]k = 16 26 66
(17.92)
The transverse shear stresses xz and yz can be determined next from the equilibrium equations (17.42) and (17.43). We obtain: k m xz = [Q1 ]k + z [Q6 ]k m + z , z x y x y k m yz = [Q 2 ]k m + z [Q6 ]k +z . z y x x y (17.93) (17.94)
The in-plane strains m and the curvatures are expressed as functions of the inplane resultants N and the moments Mf considering Expression (14.44):
m = AN + BM f , = CN + DM f .
(17.95) (17.96)
By substituting these expressions into Relations (17.93) and (17.94), the transverse shear stresses are written as:
k xz z N x = [Q1 ]k [ A + z C B + z D] M f x N y = [Q 2 ]k [ A + z C B + z D] M f y N y [Q6 ]k [ A + z C B + z D] M f y N [Q6 ] [ A + z C B + z D] x k M f x ,
(17.97) k yz z ,
(17.98) where the matrices [ A + z C B + z D] are matrices with three rows and six columns.
17.4.3.2 Cylindrical Bending

In the case of a cylindrical bending about the y direction (Chapter 19), the resultants and the moments are functions only of x, and Equations (13.56) for plates reduce to: M xy M x (17.99) = Qx , = Qy , x x which leads to the matrix equation:
358
N x M f x
on introducing the matrix:
Qx = Fx , Q y
0 0 0 . 0 0 1
(17.100)
0 0 0 Fx = 1 0 0
(17.101)
In the case of a cylindrical bending about the y direction, Expression (17.97) of k the transverse shear stress xz therefore reduces to:
Qx k xz = [Q1 ]k [ A + z C B + z D] Fx , z Q y
(17.102)
Similarly, in the case of a cylindrical bending about the x direction, Equations (13.56) for plates reduce to:
M xy y = Qx , M y y = Qy ,
(17.103)
whence the matrix relation:
N y M f y
Q = Fy x , Q y
0 0 0 . 0 1 0
(17.104)
0 0 0 Fy = 0 0 1
(17.105)
The transverse shear stress (17.98) in the layer k is written as:

Qx k yz = [Q 2 ]k [ A + z C B + z D] Fy . z Q y
(17.106)
359
17.4.3.3 Integration through the Laminate Thickness

The integration of Expressions (17.102) and (17.106), through the thickness of the laminate, leads to: Qx z2 z2 k xz = [C xz ]k [Q1 ]k z A + C z B + D Fx , (17.107) 2 2 Q y
k yz = Cyz k [Q 2 ]k z A + 2 C z B + 2 D Fy Q . (17.108) y
z2
z2
Qx
The constants of integration [C xz ]k and C yz k in each layer are determined by

k k considering the continuity of xz and yz from one layer to another, and the k k vanishing of xz on the two outer faces of the laminate. and yz
On the lower face, k = 1, z = h0 = h 2, xz = 0, thus:
1 xz ( h0 ) = 0 = [C xz ]1 [Q1 ]1 h0 A +

Whence
2 h0 h2 Qx C h0B + 0 D Fx . 2 2 Q y
[C xz ]1 = [Q1 ]1 h0 A +
and
1 xz = [Q1 ]1 ( z h0 ) A +
2 h0 h2 C h0B + 0 D , 2 2
1 2 2 z h0 C 2
( z h0 ) B +
1 2 Qx 2 z h0 D Fx . 2 Q y
At the boundary between layers 1 and 2, z = h1 and the stresses are written as:
1 xz (h1 ) = [Q1 ]1 ( h1 h0 ) A + Q x Fx , Q y
1 2 2 h1 h0 C 2
( h1 h0 ) B +
1 2 2 h1 h0 D 2
2 xz (h1 ) = [C xz ]2 [Q1 ]2 h1A +
2 h1 h2 Qx C h1B + 1 D Fx . 2 2 Q y
2 The continuity 1 xz ( h1 ) = xz ( h1 ) implies:
[C xz ]2 = [Q1 ]1 h0 A +
2 h0 h2 C h0B + 0 D 2 2 h2 h2 + ([Q1 ]2 [Q1 ]1 ) h1A + 1 C h1B + 1 D . 2 2
360
This latter expression shows that quite generally:
[C xz ]k
2 2 h0 h0 = [Q1 ]1 h0 A + C h0B + D 2 2
h2 h2 + ([Q1 ]i [Q1 ]i 1 ) hi 1A + i 1 C hi 1B + i 1 D . 2 2 i =2
(17.109)
Similarly:
2 2 h0 h0 = + + C Q A C B D h h 0 yz k [ 2 ]1 0 2 2
h2 h2 + ([Q 2 ]i [Q 2 ]i 1 ) hi 1A + i 1 C hi 1B + i 1 D . 2 2 i =2
(17.110)
The matrix for the transverse shear stresses is thus expressed as function of the transverse shear resultants in the form:
k yz Q y k = [g k ] , Qx xz
(17.111)

k g 44 [g k ] = k g54 k g 45 , k g55
(17.112)
expressed according to (17.107) and (17.108) as follows: k g 44

k g54
k g 45 = [C xz ]k [Q 2 ]k
z2 z2 z z + + A C B D Fy , (17.113) 2 2
z2 z2 k g55 z z = + + C Q A C B D Fx , (17.114) [ ] [ ] 1 k xz k 2 2
k k where the constants C xz and C yz are expressed by (17.109) and (17.110).
17.4.3.4 Estimation of the Shear Correction Factors

The strain energy per unit volume resulting from transverse shear strain is given by Expression (17.77). The strain energy per unit area of the laminate is therefore written as: U dc 1 = 2 k =1
k k k k xz + yz yz ) d z . h ( xz
k 1
hk
(17.115)
361
In layer k, the transverse shear strains are related to the transverse shear stresses by the elasticity relation:
k yz S44 k = xz S45 k yz S45 k , S55 xz
(17.116)
are the compliance constants of layer k, referred to the where the terms Sij reference directions of the laminate. Their expressions as functions of the compliance constants referred to the material directions of layer k are given in Table 11.4. The strain energy (17.115) can then be written in the following matrix form: U dc 1 = 2 k =1
S k k 44 yz xz S hk 1 45
hk hk
k 1
k yz S45 k dz , S55 xz
(17.117)
or considering (17.111) as:

n 1 U dc = Q t 2 k =1
[g k ]t
S44
S45
S45 [g k ] d z Q , S55
(17.118)
where Q t is the matrix Qy Qx , the transpose matrix of the matrix Q of the transverse shear resultants. Calculation of the strain energy is also expressed as a function of the mean strains as: 1 0 Q y 0 yz + Qx xz , 2 or, taking (17.54) into account, it results: 1 K 44 K 45 Q. U dc = Q t 2 K 45 K55 U dc = Hence, the expression for the matrix Kij : K 44 K 45 K 45 = K55
(17.119)
(17.120)
h
k =1
hk
k 1
[ g k ]t
S44 S45
S 45 [g k ] d z . S55
(17.121)
Expanding the right-hand side, we obtain: K 44 = K 45 = K55 =
h
k =1 n
hk
k 1
k k 2 k k k k k 2 S44 ( g 44 ) + 2S45 g 44 g54 + S55 ( g54 ) dz ,
(17.122)
h
k =1 n
hk
k k k k k k k k k k k S44 g 44 g 45 + S45 g 44 g55 + g54 g 45 + S55 g54 g55 d z , (17.123) k 1 k k 2 k k k k k 2 S44 ( g 45 ) + 2S45 g 45 g55 + S55 ( g55 ) dz ,
h
k =1
hk
k 1
(17.124)
k of layer k are denoted Sij where the coefficients Sij .
362
When the parameters Kij have been determined, the transverse shear coefficients Hij introduced in (17.52) are deduced considering Relations (17.55), and then the correction factors considering Relation (17.53).
17.5 CONCLUSIONS ON THE LAMINATE THEORY WITH TRANSVERSE SHEAR

The laminate theories, which consider the effects of the transverse shear deformation, differ by the expressions of the transverse shear stiffnesses Hij operating in Equation (17.52). In fact, these theories can be formulated in a same form by introducing the correction factors defined in (17.53). In the case of the initial theory (Sections 17.2 and 17.3), we have: kij = 1, i, j = 4, 5 . (17.125)
Other values (17.72) and (17.83), obtained in the case of isotropic homogeneous plates, and then in the case of orthotropic plates (Subsection 17.4.2), are also applied to the case of laminated plates: 2 kij = , 3 5 kij = , 6 i, j = 4, 5 , i, j = 4, 5 . (17.126) (17.27)
More generally (Subsection 17.4.3), the parameters kij can be evaluated by considering cylindrical bending respectively around the directions x and y. In this case, the parameters are derived, from (17.53) and (17.55), as:
k44 = k55 K55 , F44 K K 44 = , F55 K k45 = K = K 45 , F45K
2 . K 44 K55 K 45
(17.128)
The coefficients Kij are then expressed by Relations (17.122) to (17.124). The parameters kij can thus be interpreted as correction factors for the initial theory (Sections 17.2 and 17.3). The equations relating to the different theories are then obtained by replacing coefficients Fij in Expressions (17.1) to (17.51) for the initial theory by the coefficients Hij = kij Fij, with the corresponding values of the correction factors kij (Relations (17.126) to (17.128)).
Exercises
363
EXERCISES
17.1 A symmetric laminate (Figure 17.2) is made of four cloth reinforced layers and two mat reinforced layers. The cloth reinforced layers, of thickness 0.7 mm, have the following mechanical characteristics:
EL = 25 GPa, GLT = 2.2 GPa, ET = 15 GPa, GTT = 2.4 GPa. GLT = 2.5 GPa,
The mat reinforced layers have a thickness of 1mm and the mechanical characteristics are: ELmat = ETmat = 8.4 GPa, LTmat = 0.40,
GLT mat = GTT mat = 2.5 GPa.
Calculate the stiffness matrix of the laminate. The laminate is subjected to a state of transverse shear with transverse shear resultants: Qx = 48 kN/m, Qy = 0. Calculate the mean transverse shear strains of the laminate, and then the transverse shear stresses in each layer of the laminate. 17.2 Between the two mat layers of the laminate of Exercise 17.1, a layer of isotropic foam is interleaved. The moduli of the foam are: Ec = 80 MPa and c = 0.40. Do Exercise 17.1 again for three different values of the thickness of the foam layer: h = 3mm, h = 10 mm and h = 30 mm. Compare the results obtained in the case of the three thicknesses, as well as the results obtained in Exercise 17.1.
cloth cloth mat mat cloth cloth
1.4 mm 1 mm
FIGURE 17.2. Laminate of Exercise 17.1.
CHAPTER 18
Theory of Sandwich Plates
18.1 INTRODUCTION
The purpose of this chapter is to establish the equations for the mechanical behaviour of sandwich plates. A sandwich material (Chapter 3) is made from a material of low density (the core) having face sheets (the skins), made of plates or laminates bonded to each of the surfaces of the core. The essential function of the core is to transfer, by transverse shear, the mechanical loads from one skin to the other. In the general case, the skins are laminates of thickness h1 (the lower skin) and of thickness h2 (the upper skin) (Figure 18.1). The thickness of the core will be denoted by h. The coordinate system is chosen in such a way that the plane (x, y) is the middle plane.
18.2 STRAIN AND STRESS FIELDS 18.2.1 Hypotheses for Sandwich Theory
The theory of sandwich plates is based on the following basic hypotheses: 1. The thickness of the core is much greater than tat of the skins: h h1, h2. 2. The in-plane displacement in the core uc and vc in the x and y directions are linear functions of the z coordinate. 3. The in-plane displacements u and v in the x and y directions are uniform through the thickness of the skins. 4. The transverse displacement w is independent of the z coordinate: the strain zz is neglected. 5. The core transmits only the transverse shear stresses xz, yz : the stresses xx, yy, xy and zz are neglected in the core.
365
h2
h
h 2
FIGURE 18.1. Notations for a sandwich plate.
h1
6. The transverse shear stresses xz and yz are neglected in the skins. Lastly, the theory considers the elasticity problems of small deformations.
18.2.2 Displacement Field

Assumption 2 implies a first-order model for the displacements in the core: ua ( x, y, z ) = u0 ( x, y ) + z x ( x, y ), va ( x, y, z ) = v0 ( x, y ) + z y ( x, y ), with (18.1)
u0 ( x, y ) = ua ( x, y, 0), v0 ( x, y ) = va ( x, y, 0). The continuity of the displacement at the core-skin interfaces, associated with assumption 3, leads to the following expressions for the in-plane displacements within the skins: lower skin: h u1 ( x, y, z ) = u0 ( x, y ) x ( x, y ), 2 (18.2) h v1 ( x, y, z ) = v0 ( x, y ) y ( x, y ), 2 upper skin: h u2 ( x, y, z ) = u0 ( x, y ) + x ( x, y ), 2 (18.3) h v 2 ( x, y, z ) = v0 ( x, y ) + y ( x, y ). 2 Assumption 4 is written:
w ( x, y, z ) = w 0 ( x, y ).
(18.4)
366
Chapter 18 Theory of Sandwich Plates
The theory of sandwich plates is thus based on the determination of five functions of displacements and rotations: u0, v0, w0, x and y, analogous to those introduced in the laminate theory which considers the transverses shear (Chapter 17).
18.2.3 Strain Field

The strain field in the lower skin is deduced from the displacement field (18.2). It is written:
1 xx = 1 yy
u1 u0 h x = , x x 2 x v v h y = 1= 0 , y y 2 y (18.5)
1 zz = 0, 1 yz = 1 xz 1 xy
v1 w w 0 + = , z y y u w w 0 = 1+ = , z x x y u v u v h = 1+ 1= 0+ 0 x+ y x y x 2 y x
1 The transverse shear strains 1 yz and yz in the skin are neglected and the strain
field is reduced to the in-plane strains that may thus be written as:
0 1 x xx xx 1 0 h yy = yy y , 1 0 2 xy xy xy
(18.6)
with notations already introduced in Chapter 17:

0 xx =
u0 , x
0 yy =
x = x , x
v0 , y y , y = y
0 xy =
xy
u0 v0 , + y x y . = x+ y x
(18.7)
Similarly, the strain field in the upper skin may be written in the form:
2 0 xx x xx 2 0 h yy = yy + y . 2 0 2 xy xy xy
(18.8)
367
The strain field in the core is deduced from the displacement field (18.1). We obtain:
c xx =
uc u0 = +z x, x x x y vc v0 c , = +z yy = y y y w c zz = 0 = 0, z va w w 0 c + = +y , yz = z y y u w w 0 c xz = a+ = + x , z x x y u v u v c xy = a + a = 0 + 0 + z x + y x y x x y
(18.9)
The strain field has the same form as the strain field (17.2) introduced in the laminate theory with transverse shear. It is the superposition of two strain fields: the in-plane and flexural strain field:
c 0 xx x xx c 0 yy = yy + z y , c 0 xy xy xy
(18.10)
the transverse shear strain field:

c yz c xz w 0 y + y . = w 0 + x x
(18.11)
18.2.4 Stress Field

The stress field in the core is deduced from assumption 5 that leads to:
c c c xx =c yy = xy = zz = 0 .
(18.12)
The core transmits only the transverse shear stresses:

c c c C45 c yz yz C44 c = c c , c C C xz 45 xz 55
(18.13)
c c are expressed as functions of the coefficients Cij where the coefficients Cij
368
referred to the material directions of the core (Table 11.3) as:

c c c = C44 C44 cos 2 + C55 sin 2 , c c c = ( C55 ) sin cos , C45 C44
(18.14)
c c c = C44 C55 sin 2 + C55 cos 2 ,
where is the angle of the material directions of the core with the reference c in the material directions are directions of the sandwich plate. The coefficients Cij themselves expressed (10.26) as functions of the shear moduli of the core in the material directions, as follows:
c c C44 = G23 , c c C55 = G13 .
(18.15)
Assumption 6 implies that the transverse shear stresses in layer k of the lower or upper skin are zero:
k k xz = yz = 0.
(18.16)
The other stresses are deduced from the strains in the skins by the relation:
k xx Q12 Q11 k Q22 yy = Q12 k Q26 xy Q16
Q16 Q26 Q66 k
i xx i yy , i xy
i = 1, 2,
(18.17)
for layer k of the lower skin (i = 1) or upper skin (i = 2).
18.3 GOVERNING EQUATIONS OF SANDWICH PLATES 18.3.1 Constitutive Equation

The constitutive equation of sandwich plates considers the resultants and moments already introduced in the laminate theory: the in-plane resultants:
Nx N = y N xy Mx M = y M xy
xx d z + yy ( h 2+ h1 ) xy
h 2
h 2+ h2 h2
xx d z , yy xy xx z d z , yy xy
(18.18)
the bending and twisting moments:
xx z d z + yy ( h 2+ h1 ) xy
h 2
h 2+ h2 h2
(18.19)
18.3 Governing Equations of Sandwich Plates
369
the transverse shear resultants:

Q y = Qx
h2 h 2
yz dz . xz
(18.20)
By substituting the expressions (18.13) to (18.17) of stresses into the preceding expressions for the resultants and moments, we obtain the constitutive equation as:
N x A11 A12 N A y 12 A22 N xy A16 A26 M x = C11 C12 M y C12 C22 M xy C16 C26 Q 0 0 y 0 Qx 0 with:
1 2 + Aij Aij = Aij ,
A16 A26 A66 C16 C26 C66 0 0
B11 B12 B16 D11 D12 D16 0 0
B12 B22 B26 D12 D22 D26 0 0
B16 B26 B66 D16 D26 D66 0 0
0 0 0 0 0 0 F44 F45
0 0 xx 0 0 yy 0 0 xy 0 x (18.21) 0 y 0 xy F45 0 yz F55 0 xz
Bij =
h 2 1 Aij Aij , 2 h 2 1 Cij Cij , 2

hk hk 1
1 2 + Cij Cij = Cij ,
(18.22)
Dij = and
1 Aij
h 2
h 2+ h1 ) h 2
) dz = ( Qij k
k =1
n1
) d z = ( Qij ) ek , ( Qij k k
k =1
n1
(18.23)
1 Cij
h 2+ h1 ) h 2+ h2
) dz = z ( Qij
k n2
h
k =1 hk
k 1
n1
hk
k 1
) dz = z ( Qij
k n2
(Qij )k ek zk ,
k =1
n1
(18.24)
2 = Aij
h2 h 2+ h2 h2
) dz = ( Qij
k
h
k =1 n2 k =1
) dz = ( Qij
k
( Qij )k ek ,
k =1
(18.25)
2 = Cij
) dz = z ( Qij
k
hk
k 1
) dz = z ( Qij
k
(Qij )k ek zk ,
k =1
n2
(18.26) (18.27)
c . Fij = hCij
370
In the preceding expressions for the stiffness coefficients, n1 and n2 are the c are the numbers of layers respectively in the lower and upper skins, and Cij transverse shear constants of the core expressed by (18.14). The constitutive equation (18.21) has a form similar to Equation (17.21), obtained in the case of the laminate theory with transverse shear. It differs from it by the presence of terms Cij, which induce an asymmetry in the stiffness matrix of the sandwich plate. As it was developed in the laminate theory with transverse shear (Section 17.4), the stiffness coefficients Fij are sometimes modified by introducing correction factors kij, and replaced by the transverse shear parameters Hij defined in the same way as in (17.53):
H ij = kij Fij .
(18.28)
In the case of symmetric sandwich plates, the lower and upper skins are identical, hence:
1 2 Aij = Aij , 1 2 Cij = Cij .
(18.29)
So, it results that:

2 Aij = 2 Aij , 2 Dij = hCij ,
Bij = 0,
Cij = 0.
(18.30)
Thus, in the case of symmetric sandwich plates, there is no coupling between inplane and flexural behaviours. The constitutive equation then takes a form identical to the constitutive equation of symmetric laminates with transverse shear.
18.3.2 Governing Equations

The governing equations of sandwich plates are derived by introducing the constitutive equation (18.21) in the fundamental relations (13.50) of plates. The three first equations are identical to Equations (17.27), (17.28) and (17.29) with Aij, Bij and Fij defined in (18.22) and (18.27). The last two equations are written as:
C11 2u0 x 2 + 2C16 2 u0 2 u0 2v0 2v0 2v0 C C C C + C66 + + + + ( ) 16 12 66 26 xy xy y 2 x 2 y 2
+ D11
2 x x 2
2 y 2 y 2 x 2 x + 2 D16 + D66 + D16 + ( D12 + D66 ) xy xy y 2 x 2
+ D26
2 y y 2
2u0 t 2
w 0 w 0 F55 x + F45 y + x y
+ I xy 2 x t 2
=R
(18.31)
18.4 Sandwich Materials with Thick Skins
371
C16
2u0 x 2
+ ( C12 + C66 )
2u0 2 u0 2v0 2v0 2v0 C C C + C26 + + + 2 66 26 22 xy xy y 2 x 2 y 2

2 2
+ D16 + D22
2 x x 2 2 y y
2
y y 2 x 2 x D 2 D + ( D12 + D66 ) + D26 + + 66 26 xy xy y 2 x 2 w w F45 x + 0 F44 y + 0 x y

+ I xy 2 y t 2
=R
2v0 t 2
(18.32)
These equations differ from Relations (17.30) and (17.31) in the substitution of coefficients Cij for the coefficients Bij. The boundary conditions are identical to the conditions introduced in Subsection 17.3.3. In the case of symmetric sandwich materials, the form of the governing equations is identical to that of symmetric laminates with transverse shear (Subsection 17.3.2). From this it results the identity of the mechanical behaviours of symmetric sandwich materials and laminates. The two behaviours differ by the expressions of coefficients Aij, Dij, Fij. The analyses of the mechanical behaviour which will be developed for symmetric laminates with transverse shear will be able to be transposed to symmetric sandwich plates.
18.4 SANDWICH MATERIALS WITH THICK SKINS

The theory of sandwich plates implies that the thickness of the skins is much less that the thickness of the core (assumption 1). In the case of thick skins, it is possible to implement the analysis of the behaviour of sandwich plates by considering the laminate theory with transverse shear. We develop this analysis in the case of a symmetric sandwich material (Figure 18.2), and in the case where the material directions of the core and skins coincide with the reference directions of the plate. The material behaviours are characterized:
for the skins:
sk by the reduced stiffness constants Qij ,
for the core:

c by the reduced stiffness constants Qij , c . by the transverse shear moduli Gij
The application of the sandwich theory leads to the following expressions for the coefficients of the constitutive equation:
372
z h1 h h1
FIGURE 18.2. Symmetric sandwich with thick skins.

S sk Aij = 2h1Qij , S Bij = 0,
1 sk S Dij = Qij ( h1 + h ) hh1, 2

S c Fij = hGkl
(18.33)
S c ). F55 = hG13
S c ( F44 = hG23 ,
Application of the laminate theory with transverses shear leads to:

sk c , Aij = 2h1Qij + hQij
(18.34)
Bij = 0 ,
2 3 1 sk 2 h1 c h , Dij = Qij h1 ( h + h1 ) + + Qij 2 3 12 sk c Fij = 2h1Gij + hGij .
(18.35) (18.36) (18.37)
Hence:
c h Qij S Aij = Aij 1 + sk 2h1 Qij
(18.38)
Dij =
S Dij
c h1 h + 4 h1 Qij h2 3 + sk 1 + , h h + h h h + h 6 ( ) Q 1 1 1 ij
(18.39)
sk h Gij S Fij = Fij 1 + 2 1 c h Gij
(18.40)
Expressions (18.38), (18.39) and (18.40) establish the relations between the stiffness coefficients of the laminate theory with transverse shear and sandwich
Exercises
373
theory. As the core is less stiff than the skins, we have:

c sk Qij Qij ,
18.41)
and the relations can be simplified as follows:

S , Aij Aij
(18.42) (18.43) (18.44)
h h1 h + 4 S 3 1 Dij Dij + 1 , h h + h1
sk h1 Gij S 1 2 Fij = Fij + c h Gij
We recover the behaviour of a sandwich material: the in-plane behaviour is determined by the skins and the transverse shear is imposed by the core. The flexural coefficients Dij are, however, modified with respect to the sandwich theory. Expression (18.43) thus allows us to evaluate the influence of the thickness of the skins. For example, in the case where
h1 = 3 mm, h = 10 mm,
we find that:
S Dij = 1,323Dij
which is a difference of more than 30 % between the two analyses.
EXERCISES
18.1 We consider the sandwich material of Exercise 17.2. For each thickness of the core, calculate the stiffness matrix by using the theory of sandwich plates. Compare to the results found in Exercise 17.2, using the laminate theory with transverse shear.
Part V
Mechanical Behaviour of Composite Structures
This part develops the elements of the analysis of bending, buckling and vibrations of structures constituted of composite materials. The simplest analyses are the ones for which the analyses can be reduced to a one-dimensional analysis. This is the case of the cylindrical bending of plates (Chapter 19) and that of the beam bending (Chapter 20). Chapter 21 studies the bending of orthotropic laminate plates for which there is no in-plane and flexural coupling. Chapter 22 considers the bending behaviour of different laminate plates: symmetric laminates, cross-ply laminates and angle-ply laminates. The analyses show the difficulties of finding analytical solutions. Buckling of beams and plates is considered in Chapter 23. Next, vibrations of beams and plates are considered in Chapter 24. The last Chapter 25 analyses the problem of predesigning a laminate or sandwich structure, establishing a general synthesis of the concepts developed throughout the book.
CHAPITRE 19
Cylindrical Bending
19.1 INTRODUCTION
In this chapter and Chapter 20, we are interested in problems for which the theory of plates can be reduced to one-dimensional analysis. The first type of problem concerns plates that have a quite length-to-width ratio so that the deformation of the plate may be considered to be independent of the coordinate along the length of the plate. Such behaviour is called cylindrical bending and is considered in the present chapter. The second type of problem is that of analysing the bending behaviour of beams, which will be the purpose of Chapter 20.
19.2 CLASSICAL LAMINATE THEORY 19.2.1 Equations

We consider a plate made of a laminate with an arbitrary number of layers that is very long in the y direction (Figure 19.1). The plate is supported along the length of its edges x = 0 and x = a. If the transverse load is a function only of x: q = q(x), the deformation of the plate is cylindrical, that is:
u0 ( x, y, t ) = u0 ( x, t ), v0 ( x, y, t ) = v0 ( x, t ), w 0 ( x, y, t ) = w 0 ( x, t ).
(19.1)
By substituting Equation (19.1) into the fundamental equations (16.1) to (16.3) of the classical laminate theory, we obtain the one-dimensional equations:
A11
A16
2 u0 x 2
2u0 x 2
+ A16
+ A66
2v0 x 2
2v0 x 2
B11
B16
3w 0 x3
3w 0 x3
= s
= s
2u0 t 2
2v0 t 2
, ,
(19.2) (19.3)
378
Chapter 19 Cylindrical Bending
a x
FIGURE 19.1. Plate of great length.
D11
4w 0 x 4
B11
3u0 x3
B16
3v0 x3
= q s
2w 0 t 2
(19.4)
In the case of a static bending, these equations can be uncoupled by expressing u0 and v0 from Equations (19.2) and (19.3) as:
d 2 u0 d x2 d 2v0 d x2
= = B d 3w 0 , A d x3 C d3w 0 , A d x3
(19.5) (19.6)
with
2 , A = A11 A66 A16
B = A66 B11 A16 B16 , C = A11B16 A16 B11.
(19.7)
By differentiating Equations (19.5) and (19.6) and substituting the results into Relation (19.4), we find the differential equation in w0:
d 4w 0 dx
4
A q, D
(19.8) (19.9)
where
D = D11 A B11B B16C .
Equation (19.8) can be integrated and the result reported in Equations (19.5) and (19.6) to obtain the equations in u0 and v0:
19.2 Classical Laminate Theory
379
d 2 u0 dx dx
2
= =
B D C D
q( x) d x , q ( x) d x .
(19.10) (19.11)
d 2v0
2
The resultants and moments are next deduced from the constitutive equation (14.29):
N x = A11
d u0 dv d 2w 0 + A16 0 B11 , dx dx d x2
d u0 d v0 d 2w 0 N y = A12 , + A26 B12 dx dx d x2 N xy = A16 d u0 dv d 2w 0 , + A66 0 B16 dx dx d x2 (19.12)
du dv d 2w 0 , M x = B11 0 + B16 0 D11 dx dx d x2 M y = B12 M xy = B16 d u0 dv d 2w 0 , + B26 0 D12 dx dx d x2 d u0 dv d 2w 0 . + B66 0 D16 dx dx d x2
19.2.2 Uniform Load

In the case of a uniform load that is independent of x: q(x) = q0, the integration of Equations (19.8), (19.10) and (19.11) leads, on taking (19.5) and (19.6) into account, to:
B x3 x2 + a1 + b1 x + c1 , q0 D 6 2 C x3 x2 v0 = q0 + a1 + b2 x + c2 , D 6 2 A x4 x3 x2 + a1 + b3 + c3 x + d3 . w 0 = q0 D 24 6 2
u0 =
(19.13)
We consider two types of supports hereafter.

1. Case of simply supports In the case of a plate simply supported along its edges x = 0 and x = a, the boundary conditions are thus for x = 0 and x = a:
N x = N xy = M x = 0 ,
(19.14)
380
w0 = 0 .
(19.15)
Relations (19.12) show that Nx, Nxy and Mx vanish at the plate supports if, for x = 0 and x = a, we have:
d u0 d v0 d 2w 0 = = = 0. dx dx d x2
(19.16)
So as to prevent rigid body displacements of the plate, we consider that the plate is fixed at the origin, which for x = 0 imposes:
u0 = v0 = 0 .
(19.17)
The set of expressions (19.13) and conditions (19.15), (19.16), (19.17) lead to the expressions of displacements:
Bq0 ( 2 x 3a ) x 2 , 12 D Cq v0 = 0 ( 2 x 3a ) x 2 , 12 D Aq0 ( 3 x 2ax 2 + a3 ) x. w0 = 24 D u0 =
(19.18)
Substituting Expressions (19.18) into Relations (19.12), the resultants and moments may be written: N x = N xy = 0, q0 ( A12 B + A26C B12 A)( x a ) x, 2D q M x = 0 ( x a ) x, 2 q M y = 0 ( B12 B + B26C D12 A )( x a ) x, 2D q M xy = 0 ( B16 B + B66C D16 A )( x a ) x. 2D Ny =
(19.19)
Relations (19.18) show that the maximum plate deflection is obtained for x = a/2, and is given by:
w 0 max =
2. Case of clamped edges
5 Aa 4 q0 . 384 D
(19.20)
In the case where the plate is clamped along the edges x = 0 and x = a, the boundary conditions, for x = 0 and x = a, are:
u0 = v0 = w 0 = 0,
d w0 = 0. dx
(19.21)
381
Taking (19.13) into account, these conditions lead to the expressions of displacements: Bq u0 = 0 ( 2 x 2 3ax + a 2 ) x, 12 D Cq (19.22) v0 = 0 ( 2 x 2 3ax + a 2 ) x, 12 D Aq0 ( 2 x 2ax + a 2 ) x 2 . w0 = 24 D By substituting these expressions into Relations (19.12), we obtain the resultants and moments: N x = N xy = 0, q0 ( A12 B + A26C B12 A) 6 x 2 6ax + a 2 , 12 D q M x = 0 6 x 2 6ax + a 2 , 12 q M y = 0 ( B12 B + B26C D12 A ) 6 x 2 6ax + a 2 , 12 D q M xy = 0 ( B16 B + B66C D16 A ) 6 x 2 6ax + a 2 . 12 D Ny =
(19.23)
( (
) )
The plate deflection is maximum at the middle of the edges: x = a/2, and by (19.22) is expressed as:
w 0 max Aa 4 q0 = . 384 D (19.24)
For the two types of boundary conditions considered, the maximum plate deflection (19.20) and (19.24) can be written in the form:
max , w 0 max = (1 + E ) w 0
(19.25) (19.26)
where E= B11B + B16C , D
max is the maximum deflection in the case where the in-plane and flexural and w 0 coupling coefficients Bij are zero. Therefore:
in the case of simply supported edges: max = w0 in the case of clamped edges: max = w0 a 4 q0 . 384 D11 (19.28) 5a 4 q0 , 384 D11 (19.27)
382
The coefficient E is always positive. From this, it results that the coupling between in-plane and flexural behaviours increases the deflection of the plate. This increase depends upon the structure of the laminate. For example, in the case of [0/90]p cross-ply laminates made of 2p identical layers, we have (Chapter 15): A16 = B16 = D16 = 0 , (19.29) and Relation (19.25) is written: w 0 max = 1 B2 1 11 A11D11 max , w0 (19.30)
with from expressions of Table 15.3: 1 E A11 = 1 + T Q11h, 2 EL 1 ET B11 = 1 Q11h 2 , 8 p EL 1 ET Q11h3 . D11 = 1 + 2 EL 12 Whence: w 0 max = 1 3 E E 1 1 2 T L 4 p 1 + ET EL
2
(19.31)
w 0max .
(19.32)
max , even for a small This expression shows that w 0 max is practically equal to w 0 number of layers. For example, in the case where ET/EL = 1/4, we obtain:
i for p = 1 (2 layers) : i for p = 2 (4 layers) :
max , w 0 max = 1.37w 0 max . w 0 max = 1.07w 0
19.2.3 Sinusoidal Load

Because every load can be expressed in the form of a Fourier series, it is interesting to solve the cylindrical bending problem in the case of a sinusoidal load. We examine the case of a simply supported plate subjected to a load of the form:
q( x) = q0 sin m x . a (19.33)
Expressions (19.8), (19.10) and (19.11), associated with the boundary conditions (19.15), (19.16) lead to:
19.3 Including the Transverse Shear Effect
383
u0 = v0 = w0 =
Ba3q0 m3 3 D Ca3q0 m3 3 D Aa 4 q0
4 4
cos m cos m sin m
x , a x , a x . a (19.34)
m D
The maximum plate deflection is written as:

w 0 max = Aa 4 q0 m 4 4 D
(19.35)
and can be put in the form:

max , w 0 max = (1 + E ) w 0
(19.36)
where E is the coupling coefficient introduced in (19.26) and

max = w0 a 4 q0 m 4 4 D11
(19.37)
is the deflection observed in the absence of coupling between the in-plane and flexural behaviours (Bij = 0). The distribution of the stresses in each layer of the laminate is derived from Relation (14.20). The transverse shear stresses can then be obtained from the fundamental relations (13.20) in the case of statics problems.
19.3 INCLUDING THE TRANSVERSE SHEAR EFFECT 19.3.1 Orthotropic Laminate

We consider the cylindrical bending (Figure 19.1) of a plate made of a laminate comprising an arbitrary number of orthotropic layers (unidirectional layers or cloth reinforced layers), the orthotropy directions of which are parallel to the x and y directions of the plate. Thus:
A16 = A26 = 0, B16 = B26 = 0, D16 = D26 = 0.
(19.38)
The plate of infinite length in the y direction is assumed to be in a state of plane deformation: u0 = u0 ( x, t ), x = x ( x, t ), v0 = 0, y = 0, w 0 = w 0 ( x, t ). (19.39) Equations (17.27) to (17.31), taking (17.53) into account, reduce to:
384
A11
2u0 x 2
+ B11
2 x x 2
= s
2u0 t 2
+R
2 x t 2
(19.40)
x 2w 0 2w 0 + k55 F55 , + q = s x x 2 t 2
B11 2 u0 x 2 + D11 2 x x 2
(19.41)
w 2u 2 x k55 F55 x + 0 = R 20 + I xy . (19.42) x t t 2
In the case of static bending, the preceding equations may be written in the form:
A11
d 2 u0 d x2
+ B11
d 2 x d x2
= 0,
(19.43)
d d 2w 0 k55 F55 x + +q = 0, dx d x2
B11
(19.44)
d 2 u0 dx
2
+ D11
d 2 x
2
dw k55 F55 x + 0 = 0 . dx dx
(19.45)
We consider the case of simply supports along the edges x = 0 and x = a. The boundary conditions, for x = 0 and x = a, are thus, by (17.35):
w 0 = 0, N x = 0, M x = 0.
(19.46)
When the constitutive equation (17.21) is taken into account, the boundary conditions, for x = 0 and x = a, are written as: w 0 = 0, du0 d + B11 x = 0, dx dx du0 d x + D11 = 0. M x = B11 dx dx N x = A11 We study the case of the sinusoidal transverse load: q( x) = q0 sin m x . a (19.48) (19.47)
The solutions satisfying the equilibrium equations (19.43) to (19.45) and the boundary conditions (19.47) are of the form: u0 = Am cos m x , a
x = Bm cos m
x , a
w 0 = Cm sin m
x . a
(19.49)
385
Substituting Expressions (19.49) into Equations (19.43) to (19.45), and then solving the system of equations obtained, we derive: Am = Bm =
2 ) m3 3 ( A11D11 B11 2 ) m3 3 ( A11D11 B11
B11a3q0
, , (19.50)
A11a3q0
1 a 2 q0 A a2 + Cm = 2 2 11 2 2. 2 m ( A11D11 B11 ) k55 F55 m The deflection is maximum at the middle of the plate (x = a/2), so:
w 0 max = Cm ,
and the deflection can be put in the form:

max , w 0 max = (1 + E + m 2 2 S ) w 0
(19.51)
max is the maximum deflection of the laminate in the absence of inwhere w 0 plane flexural bending and in the absence of transverse shear deformation. Coefficient E is the term owed to in-plane flexural coupling and S is the term that takes account of the transverse shear. These terms are expressed as follows:
max = w0 E=
a 4 q0 m 4 4 D11
2 B11
, ,
2 A11D11 B11 D11 S= . k55 F55 a 2
(19.52)
Expression (19.51) shows that the transverse shear increases the maximum deflection. The preceding expressions can also be rewritten by introducing an effective bending stiffness: Q11 , having the dimension of a modulus, and an average shear modulus: G13 , expressed as:
Q11 = 12 D11 h
3
G13 =
F55 . h
(19.53)
The maximum deflection in the absence of in-plane flexural coupling and in the absence of the transverse shear effect is written:
max = w0
12a 4 q0
m 4 4 h3Q11
(19.54)
and the shear factor is:
386
2
S=
1 Q11 h . 12k55 G13 a
(19.55)
The importance of the shear effect thus depends on the modulus ratio Q11 G13 and the ratio a/h, the ratio of the distance between the supports to the thickness of the laminate.
k k k , yy , xy , in each layer k of the The distribution of the in-plane stresses xx
laminate is derived from Relation (17.16). The transverse shear stresses are next deduced from the fundamental relations (13.20), expressed in the case of a statics problem.
19.3.2 Antisymmetric Angle-Ply Laminate

We consider the case of an antisymmetric angle-ply laminate (Chapter 15), constituted of layers oriented alternatively at , with respect to the x direction of the plate. For these laminates, we have (Relations (15.26)):
A16 = A26 = 0, D16 = D26 = 0, B11 = B12 = B22 = B66 = 0, F45 = 0.
(19.56)
The plate, which is infinitely long in the y direction, is supported along its edges x = 0 and x = a. The plate is subjected to a load q(x). Under these conditions, the deformation is cylindrical of the form:
u0 = u0 ( x), x = x ( x), v0 = v0 ( x), y = y ( x), w 0 = w 0 ( x). (19.57)
The fundamental relations (17.27) to (17.31) may here be written as:

A11 A66 d 2u0 d x2 d 2v0 d x2 + B16 + B16 d 2 y d x2 d 2 x = 0,
= 0, d x2 d d 2w 0 F55 x + + q = 0, dx d x2 B16 d 2v0 d x2 + D11 B16 d 2 x dw 0 + F = 0, 55 x dx d x2 + D66 d 2 y d x2 F44 y = 0.
(19.58)
d 2u0 d x2
In the case of simply supports along the edges x = 0 and x = a, the boundary conditions can be written in the form:
w 0 = N x = N xy = M x = M xy = 0 .
(19.59)
387
The resultants and moments are related to the functions u0, v0, x and y by the constitutive equation (17.21). So here: u0 x B16 v0 B26 y 0 0 . 0 x 0 x y D66 y 0
N x A11 N y A12 N xy 0 = M x 0 My 0 M xy B16
A12 A22 0 0 0 B26
0 0 A66 B16 B26 0
0 0 B16 D11 D12 0
0 0 B26 D12 D22 0
(19.60)
In the case where the plate is subjected to a sinusoidal load: q( x) = q0 sin m x , a (19.61)
the solutions of the equilibrium equations (19.58) and of the boundary conditions (19.59) are of the form: x x u0 = Am cos m , v0 = Bm cos m , a a x x x = Cm cos m , y = Dm cos m , (19.62) a a x w 0 = Em sin m . a The constants Am, Bm, Cm , Dm and Em are determined by substituting Expressions (19.62) into Equations (19.58). Whence the system of equations: A11 Am + B16 Dm = 0, A66 Bm + B16Cm = 0, Cm + m 1 a Em = q0 , a F55 m (19.63)
a2 a B16 Bm + D11 + 2 2 F55 Cm + F55 Em = 0, m m a2 B16 Am + D66 + 2 2 F44 Dm = 0. m Solving this system of equations allows us to determine the constants Am, Bm, Cm, Dm and Em. The distribution of the stresses inside the laminate is then deduced from Relations (17.16) and (13.20).
388
19.4 EXACT SOLUTION

An exact solution has been derived by Pagano [24], in the case of cylindrical bending of a laminate constituted of orthotropic layers, the orthotropy directions of which are parallel to the x and y directions of the laminate plate. The cylindrical bending is characterized by a plane strain state that has no strain in the y direction y, so:
yy = 0,
xy = 0,
yz = 0.
(19.64)
Furthermore, the stresses, strains and displacements are functions of only x and z. As the layers are orthotropic with material directions the same as the plate directions, the stresses in a layer are given by:
xx C11 C12 yy C12 C22 zz C13 C23 = 0 yz 0 xz 0 0 0 xy 0 C13 C23 C33
0 0 0
C44
0 0 0 0
C55
0 0 0
0 0
0 xx 0 0 0 zz , 0 0 0 xz C66 0
(19.65)
Whence
xx = C11 xx + C13 zz , yy = C12 xx + C23 zz , zz = C13 xx + C33 zz , yz = 0, xz = C55 xz , xy = 0.

From this we deduce: (19.66)
yz = 0,
xy = 0.
(19.67)
The stress-strain relations, expressed in terms of the compliance constants, are thus written: xx S11 0 S 12 zz S13 = 0 0 xz 0 0 0 S12 S22 S23 0 0 0 S13 S23 S33 0 0 0 0 0 0 S44 0 0 0 0 0 0 S55 0 0 xx 0 yy 0 zz . 0 0 0 xz S66 0
19.4 Exact Solution
389
So:
xx = S11 xx + S12 yy + S13 zz ,

0 = S12 xx + S22 yy + S23 zz ,
zz = S13 xx + S23 yy + S33 zz , xz = S55 xz .

The stress yy is expressed as a function of stresses xx and zz as: 1 ( S12 xx + S23 zz ) . S22 R13 R33 0 0 xx 0 zz , R55 xz
(19.68)
yy =
(19.69)
By substituting this expression into Relations (19.68), we obtain: xx R11 = R zz 13 xz 0

2 S12 , S 22
(19.70)
where Rij are the reduced compliance constants, expressed as:

R11 = S11 R13 = S13 R55 = S55 . S12 S23 , S 22
2 S 23 , R33 = S33 S 22
(19.71)
The equilibrium equations (13.20) reduce to:

xx xz + = 0, x z zz xz + = 0, z x and the compatibility equations (8.21) reduce to:
2 zz x 2 + 2 xx z 2 = 2 xz . xz
(19.72)
(19.73)
We consider the case of a plate subjected to a transverse load q(x) on its upper face, and simply supported along the edges x = 0 and x = a. The boundary conditions for the stresses on the upper and lower faces are thus:
zz ( x, z = h 2) = q( x), xz ( x, z = h 2) = 0,
xx (0, z ) = 0,
w (0, z ) = 0,
zz ( x, z = h 2) = 0, xz ( x, z = h 2) = 0,
xx (a, z ) = 0,
w (a, z ) = 0.
(19.74)
whereas the support conditions can be described by: (19.75)
390
Moreover, the continuity of the stresses and displacements must be satisfied between each layer:
k k +1 zz ( x, hk ) = zz ( x, hk ), k k +1 xz ( x, hk ) = xz ( x, hk ),
u k ( x, hk ) = u k +1 ( x, hk ),
w k ( x, hk ) = w k +1 ( x, hk ).
(19.76)
Because every load q(x) can be expressed as a Fourier series, we study the case where: x q( x) = q0 sin m . (19.77) a The solution of the problem to be solved is investigated by expressing the stresses in layer k in a form that satisfies the equilibrium equations (19.72), that is:
k = f k( z ) sin m xx k = zz k xz
x , a x , a
m 2 2
2
a m x = f k ( z ) cos m , a a
f k ( z ) sin m
(19.78)
where fk(z) is a function of the z coordinate to be determined (the origin of z being in the middle plane of the laminate) and where the primes indicate differentiation with respect to z. The strain field in the layer k is then deduced considering (19.70). So, we obtain:
k k k xx f k( z ) R13 = R11
m 2 2 a
2
x f k ( z ) sin m , a (19.79)
k zz k xz
2 2 k x k m = R13 f k ( z ) R33 2 f k ( z ) sin m , a a x k m = R55 f k ( z ) cos m . a a
By substituting these expressions into the compatibility equations (19.73), we obtain a differential equation in fk(z) :
k '''' k k ) + R55 R11 f k ( z ) ( 2 R13
m 2 2 a2
k f k( z ) + R33
m 4 4 a4
fk ( z) = 0 .
(19.80)
This fourth-order differential equation has the solution:

fk ( z) =
Aik exp ( rik z ),

i =1
hk 1 z hk , k = 1, 2, . . . , n, n : number of layers,
(19.81)
19.5 Comparison between the Different Theories
391
where the constants rik are given by:

r1k m = r2 k a r3k m = r4 k a ak + bk , ck ak bk , ck
(19.82)
with
k k ak = R55 + 2 R13 , k k 2 bk = ak 4 R11 R33 , k ck = 2 R11 ,
(19.83)
and where the constants Aik are to be determined. The stresses in layer k are written:
k xx
x = sin m a m 2 2 a2
Aik rik2 exp ( rik z ) ,

i =1
k zz =
sin m
x a
Aik exp ( rik z ) ,

i =1
(19.84)
k xz =
m x cos m a a
Aik rik exp ( rik z ) ,

i =1
and the displacements are given by: a x u = cos m m a

k
i =1
k m 2 2 k 2 Aik R13 2 R11 rik exp ( rik z ) , a
x w k = sin m a
i =1
k k m 2 2 R33 Aik R13 rik 2 exp ( rik z ) . rik a
(19.85)
The conditions (19.76) of continuity between each layer, associated to (19.74) on the lower and upper faces, lead to a system of 4n equations, the solution of which allows us to find the 4n constants Aik.
19.5 COMPARISON BETWEEN THE DIFFERENT THEORIES

We compare hereafter the results derived from the different theories: classical laminate theory, laminate theory including the transverse shear effect and the exact solution, in the case of a laminate made of orthotropic layers with
392
orthotropy directions parallel to the edges of the plate. The transverse load applied to the laminate is sinusoidal of the form: q( x) = q0 sin x . a (19.86)
In the case of the classical laminate theory, the maximum deflection of the plate is given by (19.36): max . w 0 max = (1 + E ) w 0 (19.87) In the case of the laminate theory that takes the transverse shear effect into account, the maximum deflection is given by Expression (19.51):
max , w 0 max = (1 + E + 2 S ) w 0
(19.88)
max , E and S are expressed in (19.52). where w 0 The calculation of the maximum deflection for the exact solution is derived from Relation (19.85). We consider hereafter two types of laminates: an antisymmetric cross-ply [0/90] laminate and a symmetric cross-ply [0/90/0] laminate.
1. [0/90] laminate
The stiffness coefficients are given by Expressions (19.31): 1 E A11 = 1 + T Q11h, 2 EL 1 E B11 = T 1 Q11h 2 , 8 EL 1 ET Q11h3 D11 = 1 + . 2 EL 12 From this, it follows by (19.53) that:
1 E Q11 = 1 + T 2 EL Q11 = 1 ET 1+ 2 EL EL , 2 ET 1 LT EL
(19.89)
(19.90)
1 G13 = ( GLT + GTT ) . 2 Substituting these expressions into (19.88), we obtain:

2 ET 1 2 EL h 1 3 EL . ET GLT + GTT 4 a 1+ E L
w 0max (1 + E ) w 0max
E 1+ T 2 EL = 1+ 12k55 1 2 ET LT EL
(19.91)
19.5 Comparison between the Different Theories
393
1.8
exact solution
max maximum deflection w 0 max (1 + E ) w 0
1.7 1.6 1.5 1.4 1.3 1.2 1.1 1.0
k55 = 2 / 3 k55 = 5 / 6 k55 = 1
transverse shear
classical theory
0.9 0 5 10 15 20
length-to-thickness ratio ( a h ) FIGURE 19.2. Maximum deflection of a [0/90] laminate as a function of the length-tothickness ratio.
max as a function of the length-toThe variations of the ratio w 0 max (1 + E ) w 0 thickness ratio a/h, obtained by the different theories, are reported in Figure 19.2 in the case of a laminate made of unidirectional carbon fibre layers, the moduli of each layer having as values: EL = 230 GPa, ET = 14 GPa, GLT = 5 GPa, (19.92) LT = 0.3. GTT = 4 GPa,
In the case of the theory with transverse shear, the deflection variations are reported for three values of k55 : 1, 5 , 2 . The curves obtained show a good 6 3
agreement between the results deduced from the theory with transverse shear and the results derived from the exact solution, the best evaluation being obtained with k55 = 5 . The classical laminate theory does not describe the effect of the length6 to-thickness ratio a/h. The results obtained by the different theories are practically similar for the large values of this ratio.
2. [0/90/0] laminate
The stiffness coefficients are, by Table 15.2, expressed as:
394
1 E A11 = 1 + T 2 EL B11 = 0,
Q11h, (19.93)
3 1 E Q h D11 = T + 7 11 . 8 EL 12
From (19.53), it follows that:

1 E Q11 = 7 + T 8 EL EL , 1 2 ET LT EL
(19.94)
1 G13 = ( GLT + GTT ) . 2
By substituting these expressions into (19.88), we obtain: E 7+ T 2 2 EL w 0 max EL h = 1+ , max 48k55 1 2 ET GLT + GTT a w0 LT EL
(19.95)
max as a function of the length-to-width since E = 0. The variation of w 0 max w 0 ratio a/h is reported in Figure 19.3 in the case of a laminate the layers of which have the characteristics given in (19.92). We also observe in this case a good agreement between the results deduced from the laminate theory including the transverse shear effect and the results derived from the exact solution.
19.6 CYLINDRICAL BENDING OF SANDWICH PLATES

We have noted in Chapter 18, the identity between the structures of the governing equations of the symmetric sandwich plate and the ones of symmetric laminates with transverse shear. The elements developed in Section 19.3 can thus be transposed to the case of cylindrical bending of sandwich plates. As an addition to these results, we consider hereafter a sandwich plate, constituted of: two identical skins the orthotropy directions of which are parallel to the x and y directions of the plate: Bij = Cij = 0, (19.96) A16 = A26 = D16 = D26 = 0, a core the material directions 1 and 2 of which are parallel to the x and y directions: F45 = 0, F44 = hG23, F55 = hG13, (19.97) where G13 and G23 are the shear moduli measured in the material directions.
19.6 Cylindrical Bending of Sandwich Plates
395
3.0
exact solution
max maximum deflection w 0 max (1 + E ) w 0
2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0 5 10 15 20 25 30 20
k55 = 2 / 3 k55 = 1
transverse shear
classical theory
length-to-thickness ratio ( a h ) FIGURE 19.3. Maximum deflection of a [0/90/90] laminate as a function of the lengthto-thickness ratio.
The deformation state of the sandwich plate is described as: u0 = 0, x = x ( x, t ), v0 = 0, y = 0, w 0 = w 0 ( x, t ). The fundamental relations (Subsection 18.3.2) here reduce to:
D11 2 x x 2 w 2 x hG13 x + 0 = I xy , x t 2
(19.98)
(19.99) (19.100)
2w 0 2w 0 + = hG13 x + q . s x x 2 t 2 Thus in the case of a static bending, we have:

D11 d 2 x
2
d w0 hG13 x + = 0, dx dx
(19.101) (19.102)
d d 2w 0 hG13 x + + q = 0. dx d x2
396
We consider the case of a uniform load: q(x) = q0, and of a plate simply supported along the edges x = 0 and x = a:
w 0 = 0,
M x = 0,
dx = 0. dx
(19.103)
Integration of Equation (19.102) leads to: d w0 hG13 x + = q0 x + C , dx and substituting this result into Equation (19.101), we obtain:
D11 d 2 x d x2 + q0 x C = 0 .
Integration of this equation, associated with condition (19.103) for the supports, leads to: d x q = 0 x ( x a) . dx 2 D11 (19.104)
Substituting this expression into Relation (19.102), we obtain the differential equation in w0: d 2w 0 dx
2
1 1 = q0 x ( x a) . hG13 2 D11
(19.105)
Integrating twice, and then taking into account that w0 is zero for x = 0 and x = a, we finally obtain:
w0 =
q0 24 D11 12 D11 x x3 2ax 2 + a3 + ( a x ) . hG13
(19.106)
Equation (19.101) can next be solved for x, whence:
x =
q0 ( 3 4 x 6ax 2 + a3 ) . 24 D11
(19.107)
The stresses in layer k of the upper or lower skin are deduced from (18.17), and are written as: h dx k k k xx Q11 Q12 0 2 dx k k k Q22 0 0 . yy = Q12 k k 0 Q66 xy 0 0
19.6 Cylindrical Bending of Sandwich Plates
397
So:
k k xx = Q11 k yy
h dx , 2 dx k h dx , = Q12 2 dx
(19.108)
k = 0. xy
A plus sign is associated to the upper skin and a minus sign to the lower skin. The transverse shear stress can next be deduced from the fundamental relations (13.20), which here reduce to:
k k xx xz + =0. x z
(19.109)
So:
k xz k h q0 = Q11 ( 2x a ) , z 2 2 D11
(19.110)
or after integration:
k xz = k Q11 h ( 2 x a )( z + ck ) q0 . 4 D11
(19.111)
The constants ck in each layer are determined by making zero the transverse shear stress xz on the upper and lower faces of the sandwich, and by assuring the continuity of xz between each layer. We consider the case of the symmetric sandwich the skins of which are constituted of [0/90] cross-play laminates with two unidirectional layers of the k of the layers are: same thickness. The coefficients Q11
0 Q11 =
EL
2 1 LT
ET EL
90 Q11 =
ET
2 1 LT
ET EL
(19.112)
In the lower skin, the transverse shear stress in the 0 layer is deduced from (19.111): EL h 0 xz = ( 2 x a )( z + c0 ) q0 . 2 ET 4 D11 1 LT EL
The constant c0 is such that: 0 xz x, z = + h1 = 0 , h 2 hence:
0 xz =
EL h 2 ET 8D11 1 LT EL
( 2 x a )( h + 2h1 + 2 z ) q0 .
(19.113)
398
The transverse shear stress in the 90 layer is:

90 xz =
ET h 2 ET 4 D11 1 LT EL
( 2 x a )( z + c90 ) q0 .
The constant c90 is obtained by considering the continuity of xz between the layers: h h1 h h1 0 90 xz x, z = + = xz x, z = + . 2 2 2 2 So: EL h ET 90 = xz (2x a) ( 2 z + h + h1 ) + 2h1 q0 . (19.114) EL 2 ET 16 D11 1 LT EL As the stress xx is zero in the core, it results from (19.109) that the stress xz in the core is independent of z. It is obtained by stating the continuity at the skin-core interface: h a 90 xz = xz x, z = . 2 Whence: E 2+ T EL EL hh1 a xz = (19.115) ( 2 x a ) q0 . 2 ET 16 D11 1 LT EL The stress variation in the upper skin is obtained by symmetry from the variation in the lower skin. Figure 19.4 reports the variation of the transverse shear stress through the thickness of the sandwich, for x = 0 and in the case where EL/ET = 4.5 and h1/h = 0.1. We may note that the assumption of uniform strain through the skins for the sandwich theory leads to a linear variation of the transverse shear stress instead of a parabolic variation in the case of the laminate theory. The transverse shear stress is maximum in the core with a value independent of the properties of the core. This maximum stress can lead to a fracture of the sandwich induced by delamination in the case where the skin-core interface is not high enough.
EXERCISES
19.1 A plate of large dimension in the y direction is subjected to a uniform load q0. The plate is supported in the y direction, the supports being a distant a apart in the x direction. We study the cylindrical bending of this plate, for two types of supports: (1) the case of two simple edges, and (2) the case of two clamped edges.
Exercises
399
0.50
h1
0 90
0.25
h h1
90 0
ht
z ht
0.25
h1 = 0.1 h
Q11 =
EL ET = 4.5
EL
2 1 LT
0.50
0.005
0.010
2
0.015
ET EL
xz D11 q0Q11 ah
FIGURE 19.4. Variation of the transverse shear stresses through the thickness of a sandwich plate.
The plate is made of a laminate of six cloth reinforced layers. Each layer of thickness 1 mm has the same mechanical characteristics:
LT = 0.15,
GLT = 2.4 GPa.
For both support conditions, derive the deflection at an arbitrary point of the plate, the deflection at the centre of the plate, the resultants and moments, the stresses in each layer.
19.2 The half of the six-layer laminate, considered in the preceding exercise, forms the skins of a symmetric sandwich material the core of which is isotropic and has a thickness of 20 mm and the following mechanical characteristics:
Ec = 80 MPa, Gc = 35 MPa.
Do the preceding exercise again in the case where the plate is constituted of this sandwich material.
CHAPTER 20
Bending of Laminate and Sandwich Beams
20.1 INTRODUCTION
The importance of developing an analysis of the bending behaviour of beams is related to the use of beams as basic elements of structures, as well as to the mechanical characterization of laminates and sandwiches using test specimens in the form of beams. Unlike the case of cylindrical bending (Chapter 19), the theory of beams considers that the length L of the beam is much greater than its width b (Figure 20.1). The difference between cylindrical bending and beam bending is analogous to the difference between plane strains and plane stresses in the theory of elasticity. In the present chapter, we study the bending of beams made of symmetric laminates or sandwiches for which there is no in-plane flexural coupling. The direction x will be taken along the beam length and the beam thickness will be denoted h. z b y
h 2
x
FIGURE 20.1. Beam element.
401
20.2 CLASSICAL LAMINATE THEORY 20.2.1 General Expressions

In the case of a pure bending of a symmetric laminate, the constitutive equation (14.29) reduces to:
M x D11 M = D y 12 D16 M xy D12 D22 D26 D16 x D26 y , D66 xy
(20.1)
where x, y and xy are defined in Relations (14.15):
x =
2w 0 x 2
( x, y ),
y =
2w 0 y 2
D12 D22 D26
( x, y ),
xy
2w 0 = 2 ( x, y ). (20.2) xy
Equation (20.1) can be written in the following inverted form as: x D11 = D y 12 D16 xy
D16 M x D26 M y , M xy D66
(20.3)
where Dij are the components of the inverse matrix of [Dij]:

D11 = D16 = D26 =
2 ( D22 D66 D26 ),
D12 = D22 = D66 =
( D16 D26 D12 D66 ) ,

2 ( D11D66 D16 ), 2 ( D11D22 D12 ),
( D12 D26 D16 D22 ) , ( D12 D16 D26 D11 ) ,
(20.4)
and is the determinant of the matrix [Dij]:

2 2 2 = D11D22 D66 + 2 D12 D16 D26 D11D26 D22 D16 D66 D12 .
The beam theory makes the assumption that, in the case of bending along the x direction, the bending and twisting moments My and Mxy are zero:
M y = 0, Relations (20.2) and (20.3) thus lead to:

2w 0 x
2
M xy = 0.
(20.5)
x =
Mx . = D11
(20.6)
Lastly, the beam theory makes the additional assumption that the beam deflection is a function of x only:
402
Chapter 20 Bending of Laminate and Sandwich Beams
FIGURE 20.2. Effect of bending-twisting coupling in the case of a beam made of laminate material.
w 0 = w 0 ( x) .
(20.7)
The greatest attention must, however, be paid to this latter hypothesis. In fact, Equations (20.2) and (20.3) show that the curvatures y and xy are functions of the bending moment Mx, that is:
y = xy
2w 0 y
2
= D12 Mx,
2w 0 = 2 = D16 M x. xy
(20.8)
These relations show that the deflection w0 cannot be strictly independent of the variable y. This effect is particularly important in the case of laboratory bending test specimens, in a shape closer to a thin plate than a beam. From this, it follows that the bending and twisting induced by the terms D12 in Equations and D16 (20.8) can cause the specimen to lift off its supports (Figure 20.2). This effect is negligible, however, in the case where the length-to width ratio (L/b) is sufficiently high. So, under assumption (20.7), Equation (20.6) is written as:
d 2w 0 dx
2 = D11 Mx .
(20.9)
It is usual to write this equation in the form: d 2w 0 d x2 on introducing: the effective bending modulus Ex of the beam:
Ex = 12
h D11 3
M , Ex I
(20.10)
(20.11)
403
the quadratic moment I of the cross-section of the beam with respect to the (x, y) plane: bh3 , (20.12) I = I xy = 12
the bending moment M :

M = bM x .
(20.13)
Taking into account the assumptions made, Equation (13.57) for the plate bending here reduces to: d2M x + q = 0. (20.14) d x2 Considering (20.9) and (20.10), this equation is written:
d 4w 0 dx
4 = D11 q,
(20.15)
or d 4w 0 dx with
4
p , Ex I
(20.16) (20.17)
p = bq .
The differential equation (20.15) in w0 has the same form as the differential equation (19.8) obtained in the case of cylindrical bending. The two equations differ in the coefficients introduced: D11 in beam bending and 1/D11 (for symmetric laminate) in the case of cylindrical bending. Furthermore, the fourth equation (13.58) is written here as:
dMx = Qx , dx or (20.18)
dM =Q, dx setting :
Q = bQx .
(20.19)
(20.20)
k Q16 x k Q26 y , k Q66 xy
The stresses in layer k of the laminate are deuced from (14.20) as:
k k k xx Q11 Q12 k k k Q22 yy = z Q12 k k k Q16 Q26 xy
(20.21)
k for layer k , for simplification, the stiffness coefficients Qij Denoting as Qij
404
referred to the plate directions. Hence:

k k k k ) Mx, yy = z ( Q12 D11 + Q22 D12 + Q26 D16 k k k k xy = z ( Q16 D11 + Q26 D12 + Q66 D16 ) M x . k k k k xx = z ( Q11 D11 + Q12 D12 + Q16 D16 ) M x ,
(20.22)
The expressions for these stresses can be rewritten, on introducing M and I, in the form: k k M xx = axx z, (20.23) I M k yy = ak z, (20.24) yy I k k M xy = axy z, (20.25) I with
k k k k a xx D11 + Q12 D12 + Q16 D16 ) = ( Q11
h3 , 12
3
ak yy
k k k h ) D11 + Q22 D12 D16 = ( Q12 + Q26
12
(20.26)
k k k k ) a xy D11 + Q26 D12 D16 = ( Q16 + Q66
h3 . 12
These expressions for the stresses are correct only at a significant distance (> h) from the edges of the beam. The preceding results are therefore applicable only to the case of beams that have quite a high ratio b/h. Moreover, in the case of beams of a homogeneous material, Relations (20.26) associated with Expressions (15.1) and (15.2) lead to axx = 1 and ayy = axy = 0. Equations (20.23) to (20.25) reduce to equations of the usual beam theory. The transverse shear stress in the layers is derived from the equilibrium equation (19.109), which yields:
k k d xz d xx k 1 dM = = axx z. dz dx I dx
Whence:
k xz =
Q k 2 axx z + ck . 2I
(20.27)
The constants ck in each layer are determined by making xz zero on the upper and lower faces, and by considering the continuity of xz between layers. In the case of a beam made of a homogeneous material, we have axx = 1 and the transverse shear stress vanishes on the lower and upper faces: xz = 0 for z = h/2. It results that:
405
xz =
Qh 2 8I
2 2 z = 3Q 1 4 z . 1 4 h 2bh h
(20.28)
The shear stress is maximum for z = 0, that is:
xz ( z = 0) = 0 =
3Q . 2bh
2
(20.29)
Relation (20.27) can then be rewritten in the form:

z k k xz 0 4 = a xx + dk , h
(20.30)
where dk are new constants to be determined by considering the continuity of xz through the thickness of the beam. For a beam made of a homogeneous material, Relation (20.30) reduces to (20.28). Thus: z k = 0 1 4 xz . h
2
(20.31)
20.2.2 Three-Point Bending

We consider (Figure 20.3) a three-point bending beam. The symmetry of the problem leads to consider only half of the beam. The bending moment is expressed by the relation: M = Px , 2 0 x L , 2 (20.32)
where P is the total load applied at the middle of the beam. Substituting this expression into (20.10) yields: d 2w 0 dx
2
Px , 2 Ex I
0 x
L . 2
(20.33)
In the case of simple supports, the boundary conditions for x = 0 are:

M = 0,
w 0 = 0.
(20.34)
Moreover, the symmetry requires that, for x = L/2: d w0 = 0. dx

Pl 2 2x w0 = x 3 48 Ex I L
(20.35)
Integration of (20.33) associated with conditions (20.34) and (20.35) leads to:
( ) .
2
(20.36)
The deflection wc at the centre of the beam (x = L/2) is:
406
y P
x L
FIGURE 20.3. Three-point bending beam.
wc =
PL3 PL3 = D11 . 48E x I 48b
(20.37)
This relation can be used to determine either the effective bending modulus of the beam or the coefficient D11 , knowing the deflection wc at the beam centre under the load P :
Ex = PL3 PL3 , = 48 I w c 4bh3w c
= D11
(20.38) (20.39)
48bw c PL3
The stresses in layer k are deduced from (20.23) to (20.25) as:

k k xx = 6axx k yy = 6a k yy k k xy = 6axy
P bh3 P bh3 P bh3 PL bh3 PL bh3 PL bh3
xz , xz , xz. (20.40)
These stresses are maximum for x = L/2, that is:

k k xx = 3axx k yy = 3a k yy k k xy = 3axy
z, z, z. (20.41)
407
In the case of a beam of an isotropic homogeneous material: axx = 1, and the normal stress is given by: 3PL (20.42) xx = 3 z . bh The maximum tensile stress is reached on the lower face (z = h/2), and is expressed as: 3PL . (20.43) xx max = 0 = 2bh 2 The stresses in layer k of a laminate can thus be written in the form: z k k xx = 2axx 0 , h z k yy = 2a k yy 0 , h z k k xy = 2axy 0 . h
(20.44)
As an example, we consider a symmetric laminate made of eight layers of the same thickness and oriented at 0, 45 and 90. For each layer, the following characteristics of a unidirectional glass fibre composite are used:
EL = 45 GPa, ET = 10 GPa, GLT = 4.5 GPa,
LT = 0.3.
(20.45)
Three stacking sequences are considered (Figure 20.4). The variation of the normal stress xx (20.44) through the laminate thickness is reported in Figure 20.5 for the three stacking sequences. The stress is referred to the maximum value 0 reached in the case of an isotropic material. For comparison, the stress variation in the case of an isotropic material is also reported. The results obtained clearly show the influence of the stacking sequence used. The maximum stress is reached in the external layer only in the case where the layers oriented at 0 are external. It results that the load of fracture will be strongly influenced by the stacking sequence used.
0 45 45 90 90 45 45 0 [0/45/45/90]s
90 45 45 0 0 45 45 90 [90/45/45/0]s
45 0 45 90 90 45 0 45 45/0/45/90]s
FIGURE 20.4. Different stacking sequences studied.
408
0 0.125
[0/45/45/90]s 90
0 0.125
[0/45/45/90]s 0 45
45
z/h
0.250 45 0.375 0 0.500 0 0.5 1.0 1.5
z/h
0.250 45 0.375 90 0.500 0 0.5
xx 0
xx 0
1.0
1.5
0 0.125
[0/45/45/90]s 90 45
z/h
0.250 0 0.375 45 0.500 0 0.5 1.0 1.5
xx 0
FIGURE 20.5. Influence of the stacking sequence upon the variation of the xx stress through the thickness of the laminates.
In the case of a three-point bending beam, the comparison of Relations (20.19) and (20.32) shows that Q = P/2. The transverse shear stress is thus given by Relation (20.30): z 2 k k xz 0 4 (20.46) = a xx + dk , h with 3P 0 = . (20.47) 4bh The influence of the layer stacking sequence upon the variation of the shear stress xz through the thickness of the laminate is illustrated in Figures 20.6 and 20.7, for the stacking sequences [0/45/45/90]s and [0/45/45/0]s studied
409
0.500 0.375 0.250 0.125
z/h
0.000 0.125 0.250 0.375 0.500 0 0.2 0.4 0.6 0.8 1.0 1.2
xz 0
FIGURE 20.6. Variation of the transverse shear stress through the thickness of the [0/45/45/90]s laminate.
0.500 0.375 0.250 0.125
z/h
0.000 0.125 0.250 0.375 0.500 0 0.2 0.4 0.6 0.8 1.0 1.2
xz 0
FIGURE 20.7. Variation of the transverse shear stress through the thickness of the [90/45/45/0]s laminate.
410
previously. In comparison, the variation of the shear stress (20.31) of an isotropic beam is also reported. The results obtained show that the maximum shear stress depends on the stacking sequence and that the variation of the shear stress differs clearly from the variation given by the usual beam theory for isotropic material.
20.2.3 Four-Point Bending

We now consider (Figure 20.8) a four-point bending beam, loaded by two loads P/2 applied at the quarters of the span length between the beam supports. This type of loading is currently used in tests for characterizing the behaviour of composite materials. The symmetry of the problem leads to consider only half of the beam. For the left half of the beam, the bending moment is given by: Px , 2 PL M = , 8 M = d 2w 0 dx dx
2
0 x
L , 4 L L x . 4 2
(20.48) (20.49)
Substituting these expressions into (20.10) yields: = = d 2w1 dx dx

2
= =
Px , 2 Ex I PL , 8Ex I
0 x
L , 4
(20.50) (20.51)
d 2w 0
2
d 2w 2
2
L L x , 4 2
y P/2 z
P/2
x L/4 L/2 L/4
FIGURE 20.8. Four-point bending beam.
411
on introducing:
w1 = w 0 , w2 = w0 ,
L , 4 L L for x . 4 2 for 0 x
Equation (20.50) obtained in the case 0 x L/4 is identical to Equation (20.33) obtained in the case of three-point bending. In the case of simple supports, the boundary conditions for x = 0 are:
M = 0,
w1 = 0.
(20.52)
The condition on the moment is satisfied by Relation (20.48). The symmetry requires that the slope of the deformed beam vanishes at the centre of the beam, so for x = L/2: dw2 = 0. (20.53) dx Lastly, the continuity of the deflection and the slope of the deformed beam must be assured for x = L/4, so: d w1 d w 2 = . (20.54) w1 = w 2 , dx dx Integration of Equations (20.50) and (20.51) leads, taking account of the conditions (20.52) to (20.54), to: 2 PL2 x w1 = (20.55) x 9 16 , L 192 Ex I
2 PL3 x x x 1 48 + 48 . w2 = L L 768 E x I
(20.56)
These expressions allow us to determine the deflection wq at the quarter point (x = L/4) and the deflection wc at the centre (x = L/2) : PL3 PL3 D11 , = 96 E x I 96b 11PL3 11PL3 D11 . = 768E x I 768b
wq = wc =
(20.57) (20.58)
These relations can be used to determine either the effective bending modulus of from the measurement of the deflections wq or wc: the beam or the coefficient D11
Ex =
Ex =
PL3 PL3 , = 96 I w q 8bh3w q
(20.59) (20.60)
11PL3 11PL3 , = 768 I w c 64bh3w c
412
and
D11 =
96bw q PL
P bh3 P bh P
3
768bw c PL3
(20.61)
The stresses in layer k are deduced from (20.23) to (20.25) as:

k k xx = 6a xx k yy = 6a k yy k k xy = 6axy
xz , xz , xz , 0 x L , 4 (20.62)
bh3
and
k k xx = a xx k yy k xy
3 PL z, 2 bh3 3 PL = ak z, yy 2 bh3 3 k PL = axy z, 2 bh3
L L x . 4 2
(20.63)
Comparison with Relations (20.41) shows that for 0 x L/4 the stresses are expressed by relations identical to those found in the three-point bending case. Furthermore, comparison between Relations (20.62) and (20.63) shows that the maximum stresses occur in the centre section of the beam for x lying between L/4 and L/2, the stresses being independent of x in this interval. As in the case of three-point bending, the maximum stresses are not necessarily reached on the outer faces. In the case of an isotropic beam (axx = 1), the normal stress is given by: 3PL L L (20.64) xx = z , x . 4 2 2bh3 The maximum tensile stress is reached on the lower face (z = h/2) and is expressed as: 3PL . (20.65) xxm = 4bh 2 The stresses in layer k of a laminate beam can thus be rewritten as follows: z k k xx = 2a xx xxm , h z k yy = 2a k (20.66) yy xxm , h z k k xy = 2axy xxm . h
413
These expressions have the same form as Relations (20.44) obtained in the threepoint bending case. The variation of xx/xxm is thus also given by Figure 20.5 in the case of the laminates studied previously in Subsection 20.2.2. Relation (20.19) associated to (20.48) and (20.49) shows that: Q= P , 2 L , 4 L L x . 4 2 0 x (20.67) (20.68)
Q = 0,
From this it results that the transverse shear stress is zero for L/4 x L/2. In contrast, the shear stress is given by Relation (20.30) with the same value of 0 as in the three-point bending case (20.47). The variation of the transverse shear stress through the laminate thickness for r 0 x L/4 is thus identical to that obtained in the case of three-point bending for 0 x L/2.
20.3
INCLUDING THE TRANSVERSE SHEAR EFFECT
20.3.1 General Equations

In this section, we study the effect of the transverse shear deformation on the bending of laminate beam. As in Section 20.2, the study is limited to the case of symmetric laminates. In the case of pure bending, the constitutive equation (17.56) of the laminate theory including transverse shear deformation reduces to:
(20.69)
and Qy H 44 Q = H x 45 with H 45 0 yz 0 , H 55 xz (20.70)
x = 0 yz
y y x , , , y = xy = x + x y y x w w 0 = 0 + y , xz = 0 + x. y x
(20.71)
The coefficients Dij have been introduced in the classical laminate theory (14.33), and the general expressions of the coefficients Hij are given in (17.53). In the
414
context of the initial theory including transverse shear, the coefficients of transverse shear Hij reduce to the coefficients Fij expressed in (17.24) : Fij =
k =1
) = ( hk hk 1 ) ( Cij
k
( Cij )k ek .
k =1
(20.72)
The equation of moments (20.69) here has a form identical to the equation of moments (20.1) of the classical laminate theory, the curvatures x, y and xy having different expressions. The equations of moments (20.69) and the equations of the transverse shear resultants (20.70) are uncoupled and can be written in the following inverted forms: x D11 D12 D16 M x = D D D M , (20.73) y 22 26 y 12 M xy D16 D26 D66 xy and
0 yz H 44 0 = xz H 45 H 45 Q y , Qx H 55
(20.74)
are the components of the inverse matrix of [Dij] given where the coefficients Dij are expressed considering the initial by Relation (20.4), and the coefficients H ij
transverse shear theory (Relation (17.21)), as:

H 44 = F44 =
F55 , F
H 45 = F45 =
F45 , F
H 55 = F55 =
F44 , F (20.75)
with
F =
2 F44 F55 F45 .
We consider the assumption that both the functions x and w0 are independent of the variable y, that is: x = x ( x), w 0 = w 0 ( x). (20.76)
0 The strains xx and xz are then given, according to (17.12) and (17.22), by the expressions: d xx = z x , (20.77) dx d w0 0 xz = x + . (20.78) dx
In the case of pure bending, the fundamental equations of plates (13.58) reduce to:
Qx Qy + + q = 0, x y
(20.79)
415
M x M xy + Qx = 0, x y M y M xy + Qy = 0. y x
(20.80) (20.81)
From Equation (20.73), we deduce:
x =
dx = D11 Mx , dx
(20.82)
a relation that gives the moment Mx. Assumption (20.5) for classical beam theory is also applied to the theory including the transverse shear effect:
M y = 0,
M xy = 0.
(20.83)
By substituting these relations into the equilibrium equation (20.81), we find that the transverse shear resultant Qy is zero: Qy = 0 . Relations (20.74) and (20.78) thus lead to:
0 xz = x +
(20.84)
d w0 = F55 Qx . dx
(20.85)
By substituting Expressions (20.82) and (20.85) into Equation (20.80) for plates, we obtain:
d 2 x d x2
D11 + d w0 = 0 . x dx F55
(20.86)
It is usual to write this latter equation by introducing the modulus Ex (20.11) of the beam, and the shear modulus Gxz of the beam expressed as:
Gxz = 1
hF55
(20.87)
Equation (20.86) is then written: d 2 x d x2 bh Gxz I Ex + d w0 = 0 . x dx (20.88)
Similarly, substituting Expression (20.85) into the plate equation (20.79), we obtain: d 2w 0 d x + + F55 q =0, (20.89) 2 dx dx an equation which can be put into the form: d 2w 0 d x2 + dx 1 + p = 0, d x hGxz (20.90)
by introducing, as for (20.17), the load: p = bq.
416
Equations (20.86) and (20.89), or (20.88) and (20.90), constitute the fundamental equations for the bending of beams constituted of symmetric laminates, including the transverse shear deformation. These relations allow us to derive the functions x and w0. The forms (20.86) and (20.90) are similar to the equations of the theory of homogeneous beams which take transverse shear into account [28]. In the case where the variation of the bending moment Mx is known, Relation (20.82) can be used in the form: dx M = D11 Mx = , dx Ex I (20.91)
where the bending moment M has been introduced in (20.13). A second equation can be obtained by substituting (20.91) into one of Relation (20.86) or (20.88):
dMx 1 d w0 = x + , dx dx F55
(20.92)
or dM d w0 = bhGxz x + . dx dx (20.93)
The in-plane stresses in layer k of the laminate may be written, by (17.16) and (20.82), in the same forms as Expressions (20.22). From this, it results that Expressions (20.23) to (20.25) obtained by the classical laminate theory are also applicable to the case where the transverse shear effect is taken into account. Similarly, the transverse shear stresses in the layers have the same forms (20.27) to (20.31) as the stresses given by the classical theory. In the case of beam bending, taking into account the transverse shear does not modify the distribution of the transverse shear stresses in the laminate.
20.3.2 Three-Point Bending

In the case of a three-point bending beam (Figure 20.3), the bending moment is expressed by Relation (20.32). Substituting this expression into Relation (20.91), we obtain:
x =
P 2 x + c, 4 Ex I
0 x
L . 2
(20.94)
The symmetry of the loading implies that u(L/2) = 0. From (17.1) it results that this condition yields: x ( L 2) = 0 . (20.95) This condition introduced in (20.94) leads to: PL2 x x = 1 4 16 Ex I L
( ) ,
2
(20.96)
417
or
PL2 x x = D11 1 4 16b L
( ) .
2
(20.97)
By substituting Expression (20.32) for the moment M into Equation (20.93), we deduce the expression of the deflection w0 as a function of x. Thus: d w0 P x + = dx 2bhGxz , 0 x L . 2 (20.98)
It is interesting to note that, by this result, the slope of the deformed shape does not vanish at the centre of the beam. In fact, since x (L/2) = 0, the slope is: d w0 P ( L 2) = . dx 2bhGxz (20.99)
After substitution of x, integration of Equation (20.98), taking into account w0(0) = 0, leads to: x w0 = x 4 4bh3 Ex L PL2 Ex h Gxz L
()
3 2S ,
(20.100)
on introducing the shear coefficient S defined by: S=
()
= 12
F55 h L D11
( ).
2
(20.101)
The effect of the transverse shear deformation thus depends on the span-tothickness ratio L/h of the beam and on the ratio Ex /Gxz of the beam moduli. The deflection at the centre of the beam is derived from (20.100) and its absolute value is written as: PL3 ( (20.102) wc = 1+ S ) , 4bh3 E x or PL3 F 1 . (20.103) wc = D11 1 + 12 55 2 48b D11 L
This latter expression shows that it is possible to derive the coefficients D11 and F55 from measurement of wc/P for two different values of the span length L between the supports. Expressions (20.101) and (20.102) show that the deflection can be written in one of the two forms: w c ( S ) = (1 + S ) w c ( 0 ) , (20.104)
w c ( S ) = 1 +
Ex h Gxz L
( ) w
2
c ( 0) ,
(20.105)
418
with
E x 12 F55 = 2 , Gxz h D11
(20.106)
where wc(S) is the deflection obtained by considering the transverse shear effect, whereas wc(0) is the deflection given by (20.37) when the transverse shear is not considered. It results that neglecting the transverse shear effect leads to underestimating the value of the deflection. As an example, we consider the case of the symmetric laminate [0/90/0], constituted of three unidirectional layers of the same thickness. The bending stiffness coefficients are deduced from Table 15.2 as: 1 E Q h3 D11 = 7 + T 11 , 8 EL 12 D22 1 ET Q11h3 = 1 + 7 , 8 EL 12 D11 = Q12 h3 , 12 D16 = 0, D66 GLT h3 = . 12 (20.107)
D26 = 0,
Relation (20.72) leads to: h F44 = F55 = ( GLT + GTT ) , 2 Using Relations (20.4), we obtain:
= D11
F45 = 0.
(20.108)
D22
2 D11D22 D12
(20.109)
Combining Expressions (20.107) and (20.109) yields:

D11 =
1
ET 2
12
2 64 LT E L 1 7 + ET 1 + 7 ET 1 2 LT EL EL
3 1 ET EL h 7 + 8 EL ET EL
(20.110)
Considering the usual values of the moduli, we have:

D11 2 1 LT ET EL
12
1 E E L h3 7+ T 8 EL
(20.111)
By noticing that F55 = 1 F55 , we obtain by substituting the expressions for D11 and F55 in Expression (20.106) the following expression for Ex /Gxz :
419
2 2 ET 64 LT Ex EL = 1 Gxz 7 + ET 1 + 7 ET EL EL
1 ET 7 + EL EL 8 , 1 2 ET 1 ( G + G ) LT LT TT EL 2
(20.112)
with
1 E 7+ T Ex 8 EL EL . Gxz 1 2 ET 1 ( G + G ) LT LT TT EL 2 (20.113)
In the case of a laminate constituted of unidirectional glass fibre layer having as moduli: EL = 45 GPa, ET = 10 GPa, GLT = 4.5 GPa, (20.114) LT = 0.30, GTT = 4 GPa. we obtain: Ex = 8.55 . (20.115) Gxz In the case of a laminate constituted of unidirectional carbon fibre layer having as moduli: EL = 230 GPa, ET = 14 GPa, GLT = 5 GPa, (20.116) LT = 0.30, GTT = 4 GPa. we have: Ex = 44.6 . (20.117) Gxz The variation of wc(S)/wc(0) as a function of the span-to-thickness ratio L/h is reported in Figure 20.9 for the two values (20.115) and (20.117) of the ratio Ex/Gxz.
20.3.3 Four-Point Bending

In the case of a four-point bending beam (Figure 20.8), the bending moment is given by Relations (20.48) and (20.49). Substituting these expressions into Relation (20.91), we obtain: P 2 x + c1 , 4 Ex I P x = 2 = x + c2 , 8Ex I
x = 1 =
L , 4 L L x . 4 2 0 x
(20.118) (20.119)
As in the case of three-point bending, the symmetry of loading implies:
2 ( L 2 ) = 0 .
(20.120)
420
E x Gxz = 8.55
3
Ex Gxz = 44.6
wc / wc(0)
1 without transverse shear 0 0 5 10 15 20
L/h
FIGURE 20.9. Variation of the deflection at the centre of the beam for three-point bending as a function of the span-to-thickness ratio.
This condition, introduced in (20.119), leads to: PL2 x 2 = 1 2 . L 16 E x I The continuity of the displacements for x = L/4 requires that:
(20.121)
1 ( L 4 ) = 2 ( L 4 ) .
The combination of Relations (20.118), (20.121) and (20.122) leads to: PL2 x 3 16 64 Ex I L
(20.122)
1 =
( ) .
2
(20.123)
Taking into account Relations (20.48) and (20.49), Expression (20.93) is written as: d w 0 d w1 P = = 1 + dx dx 2bhGxz d w0 d w2 = = 2 , dx dx , 0 x L , 4 (20.124) (20.125)
L L x . 4 2
421
Integration of Relation (20.124), after substitution of Equation (20.123) and taking into account the condition: w1(0) = 0, leads to:
w1 =
PL2 192 Ex I
x x 16 L
( ) 9 8S ,
2
(20.126)
where S has been introduced in (20.101). Similarly, Equations (20.121) and (20.125) lead to:
w2 =
PL2 x x 1 + c3 . 16 Ex I L
( )
(20.127)
The constant c3 is determined by considering the continuity of the deflection at L/4: w1 ( L 4 ) = w 2 ( L 4 ) . (20.128) We thus obtain: PL3 x x 2 w2 = 1 48 48 (20.129) + 8S . 768E x I L L
()
The deflection at the beam centre is deduced from the previous expression and may be written as: PL3 ( wc = 11 + 8S ) , 768Ex I or
wc =
(20.130)
PL3 F 1 . D11 11 + 8 55 2 768b D11 L
(20.131)
This latter expression shows that, as in the case of three-point bending, it is and F55 from measurement of wc/P for two possible to derive the coefficients D11 different values of the span length L between the supports. Expression (20.130) shows that:
wc (S ) = 1 +
8 S w c (0) , 11
(20.132)
or
8 Ex h w c ( S ) = 1 + 11 Gxz L
( ) w (0) ,
2 c
(20.133)
where wc(0) is the deflection (20.58) neglecting the transverse shear effect. Relation (20.132) shows that considering the transverse shear effect increases the deflection at the beam centre, to a less value than in the case of three-point bending. The variation of wc(S) /wc(0) as a function of the span-to-thickness ratio L/h is reported in Figure 20.10, for the same values of the ratio Ex/Gxz considered in Figure 20.9.
422
E x Gxz = 8.55
3
Ex Gxz = 44.6
wc / wc(0)
1 without transverse shear 0 0 5 10 15 20
L/h
FIGURE 20.10. Variation of the deflection at the centre of the beam for four-point bending as a function of the span-to-thickness ratio.
20.4 BENDING OF SANDWICH BEAMS 20.4.1 General Expressions

The similarity of behaviour between the symmetric sandwich plates and the symmetric laminates with transverse shear effects included allow us to transpose the results obtained in Subsections 20.3.1 to 20.3.3 to the bending of sandwich beams. In fact, in the case of pure bending, the constitutive equation (18.21) of sandwich materials reduces to:
(20.134)
and Qy H 44 Q = H x 45 H 45 c yz , c H 55 xz (20.135)
20.4 Bending of Sandwich Beams
423
with
x = a yz
y y x , , . y = xy = x + x y y x w w a = 0 + y , xz = 0 + x. y x
(20.136)
The general expressions for the coefficients Dij and Fij are given by Relations (18.22) to (18.27). Comparing Expressions (20.134) and (20.135) with Relations (20.69) and (20.70) confirms the possibility of transposing to the bending of sandwich beams the results obtained in Subsections 20.3.1 to 20.3.3 for the functions x, y and w0. In addition to the differences between the expressions for the coefficients Dij and Fij, the essential difference between the two types of materials lies at the level of the stress. To illustrate this aspect, we consider again the symmetric sandwich studied in Section 19.6: two identical skins with orthotropy directions parallel to the directions x and y of the beam, and core with material directions 1 and 2 parallel to the directions x and y. The in-plane stresses in layer k of the upper or lower skin are given by Relations (18.17). Whence:
k k xx = Q11 k yy
h dx , 2 dx k h dx = Q12 , 2 dx
(20.137)
k = 0, xy
a plus sign plus being associated to the upper skin and a minus sign with the lower. Applying this result to the case of three-point bending where the function x is given by Relation (20.96), we obtain:
k k xx = Q11 k yy
Ph Ph k x= D11Q11x, 4 Ex I 4b Ph k k Ph x= D11Q12 x, = Q12 4 Ex I 4b
(20.138)
k xy = 0.
k The stresses are maximum for x = L/2. In particular the normal stress xx may be written as: hh 2 k k (20.139) Q11 , xx = 0 t D11 12
where the expression for 0 was introduced in (20.43). The transverse shear stress can then be deduced from the equilibrium equation (19.109), which yields:
424
k xz Ph k k Ph = Q11 = D11Q11 . z 4 Ex I 4b
(20.140)
So upon integration:
k k D11 xz = Q11
Ph ( z + ck ) , 4b
(20.141)
or
k k D11 = 0Q11 xz
hht2 z 2 + dk , 6 ht
(20.142)
on introducing the stress 0 defined in (20.47). The constants ck or dk are determined by setting xz to zero on the upper and lower faces, and by ensuring the
c , which is continuity of xz from layer to layer. The transverse shear stress xz constant through the thickness of the core, can be derived from (20.141) or (20.142) from the continuity at the skin-core interface.
20.4.2 Comparison between Sandwich Theory and Laminate Theory with Transverse Shear
20.4.2.1 Stiffness Coefficients
In such a way to compare the results deduced from the sandwich theory and laminate theory with transverse shear, we study the case of a sandwich beam with thick skins, for three-point bending of the beam. The sandwich material is made of two mat layers of thickness h1 and an isotropic core of thickness h. The skins are characterized by their moduli (15.81) and (15.82):
E Lm , GLTm =
LTm ,
E Lm , 2 (1 + LTm )
ETm = ELm , GTT m = Gm , (20.143)
and the reduced stiffness matrix of the skins is written as:
LTm 1 m m 1 Qij = Q11 LTm 0 0

with
m Q11 =
0 , 1 LTm 2
(20.144)
E Lm
2 1 LT m
(20.145)
The isotropic core is characterized by its Youngs modulus Ec and its Poisson ratio c, the shear modulus being deduced from them by the relation:
425
Gc =
Ec . 2 (1 + c )
(20.146)
The reduced stiffness matrix of the core is thus written as:

c Qij c = Q11 c
c
1 0
Ec
0 0 , 1 c 2
(20.147)
with
c Q11 = 2 1 c
(20.148)
Assuming that the core is very much less stiff than the skins, the bending stiffness coefficients of the laminate theory with transverse shear and sandwich theory are related by Expression (18.43). Whence:
S , Dij D Dij
(20.149)
with
D = 1+
h h1 h + 4 3 1 . h h + h1
(20.150)
From this, we deduce that the relation between the inverse coefficients of the two theories may be written by (20.4) as:
Dij =
S Dij .
(20.151)
S The coefficients Dij are derived from Relations (18.26) and (18.30). Thus: S 2 Dij = hCij ,
(20.152) (20.153)
with
2 Cij =
h 2+ h1 h2
1 m m Qij z d z = Qij ( h + h1 ) h1 . 2
Whence 1 m S Dij = Qij ( h + h1 ) hh1 . 2 The bending stiffness matrix is thus written as: LTm 1 S S 1 Dij = D11 LTm 0 0 with
0 , 1 LTm 2
(20.154)
(20.155)
426
1 m 1 E Lm S D11 = Qij ( h + h1) hh1 = ( h + h1) hh1 . 2 2 2 1 LT m
(20.156)
S of the inverse matrix inverse are given by Relations (20.4), The coefficients Dij
thus here:
S D11 S S Dij = D12 0 S D12 S D22
0 0 , S D66
(20.157)
with
S D11 = S S D22 = D11 ,
2 , ( h + h1 ) hh1ELm
(20.158)
S S D12 = LTm D11 ,
S S D66 = 2 (1 + LTm ) D11 . (20.159)
From Relations (18.33) and (18.44), the transverse shear stiffnesses are expressed as: h1 Gm S (20.160) Fij = Fij 1 + 2 h G , c with S S S F44 = F55 = hGc , F45 = 0. (20.161)
20.4.2.2 Deflection
The introduction of Expressions (20.158) and (20.161) in Relation (20.103) leads to an expression of the deflection wc at the beam centre given by the sandwich theory in the form:
S wc =
PL3 24b ( h + h1 ) hh1ELm
ELm ( h + h1 ) h1 1 + 6 G . L2 c
(20.162)
Similarly, on substituting the coefficients Dij given by (20.151) into Expression (20.103), the laminate theory with transverse shear leads to the following expression for the deflection at the beam centre:
wc =
6 ( h + h1 ) h1 PL3 E Lm . (20.163) 1+ D 2 h1 Gm 24b ( h + h1 ) hh1ELm L Gc 1 + 2 h Gc
In the case of high span lengths between the supports, the two expressions are related by: 1 S wc = wc . (20.164)
The sandwich theory increases the deflection derived from the laminate theory
427
with transverse shear. For example in the case where:

h1 = 3 mm, h = 10 mm,
we have D = 1.323, so there is a difference of more than 30 % between the two analyses.
20.4.2.3 Stress Distribution

In the context of the laminate theory (with or without transverse shear), the normal stresses xx in the skins for x = L/2 are deduced from (20.44) and (20.26). We obtain: z m m xx = 2axx 0 , (20.165) ht with ht = h + 2h1 , (20.166) and
m axx m m ht D11 + Q12 D12 ) = ( Q11 3
12
(20.167)
So, considering Relations (20.145), (20.151), (20.158) and (20.159), we obtain:

m axx =
ht3 . 6 ( h + h1) hh1 z , ht
(20.168)
The stress xx in the core is similarly given by:

c c xx = 2axx 0
(20.169)
with
c = a xx
1 c LTm Ec m axx . 2 E Lm 1 c
(20.170)
The variation of the stresses is given in Figure 20.11a (solid lines) in the case:
h1 = 3 mm,
h = 10 mm,
c = 0.40,
LTm = 0.30,
E Lm = 30. Ec
(20.171)
For comparison, the variation of the stress (20.42) in the case of a homogeneous beam is also reported in Figure 20.11a (dashed line). The distribution of the transverse shear stresses according to the laminate theory is given by Expression (20.46) :
k xz
k a xx 0
z 2 4 + dk , ht
(20.172)
where the constants dk are determined so as to ensure the vanishing of the stresses
428
8 6 4
8 6 4
z ( mm )
0 2 4 6 8 1.4 1.0 0.0 1.0 1.4 (a)
z ( mm )
2 0 2 4 6 8 1.4 1.0 0.0 1.0 1.4 (b)
xx 0
xx 0
FIGURE 20.11. Variation of stress xx through the thickness of the sandwich material, in the case of (a) laminate theory with transverse shear and (b) sandwich theory.
on the upper and lower faces, and the continuity at the skin-core inter-face. Whence: z 2 m m xz = axx 0 1 4 , (20.173) ht
c m c = xz xz ( h 2 ) + axx 0 2 h 2 z . 1 4 h ht t
(20.174)
The variation of the transverse shear stress is given in Figure 20.12 (solid curve) with the numerical values of (20.171). For comparison, the variation of the shear stress (20.31) for a homogeneous beam is also reported (dashed curve). According to the sandwich theory, the normal stresses in the skins for x = L/2 are deduced from (20.139) as:
m xx = 0
hht2 S m D11 Q11 . 12
(20.175)
So, from Relations (20.145) and (20.156) we deduce:

m xx = 0
ht2 1 . 2 6 ( h + h1 ) h1 1 LT m
(20.176)
The stress variation is reported in Figure 20.11a. The stress is constant in the skins and zero in the core.
Exercises
429
8 6 4
z ( mm)
2 0 2 4 6 8 0 0.2 0.4 0.6 0.8 1.0
xz 0
FIGURE 20.12. Variation of the transverse shear stress through the thickness of a sandwich material, deduced from the laminate theory.
The transverse shear stresses in the skins at x = L/2 are given by Relation (20.142), which leads to: 1 ht2 1 + 2 z , m (20.177) = 0 xz 2 ht 1 LTm 3 ( h + h1) h1 in the lower skin. The shear stress in the core is constant and equal to the stress at the skin-core interface. Whence: 2 1 ht c xz = 0 . (20.178) 2 3 1 LTm h + h1 The variation of the transverse shear stress is then given in Figure 20.13 (solid curve) for the numerical values (20.171) and compared with the case of a homogeneous beam (dashed curve).
EXERCISES
20.1 A beam made of a symmetric laminate is subjected to a uniform linear transverse load p0: q(x) = p0/b.
20.1.1 Express the deflection at an arbitrary point of the beam, the deflection at the beam centre, the bending moment, the in-plane and transverse shear stresses in layer k. The expressions introduce four constants of integration which depend on the conditions at the two ends of the beam.
430
8 6 4
z ( mm)
2 0 2 4 6 8 0 0.2 0.4 0.6 0.8 1.0
xz 0
FIGURE 20.12. Variation of the transverse shear stress through the thickness of a sandwich material, deduced from the sandwich theory.
20.1.2 Apply the preceding results to the case of a beam clamped at both ends. Then, derive the results in the case of a beam made of a symmetric laminate of five layers. Layers 1, 3, 5 are mat reinforced layers with the characteristics:
EL = ET = 8.5 GPa,
LT = 0.30,
GLT = 3.2 GPa.
Layers 1 and 5 have a thickness of 1 mm. Layer 3 is double thickness, at 2 mm. Layers 2 and 4 are unidirectional layers of thicknesses 1.6 and with the characteristics:
LT = 0.30,
GLT = 5.2 GPa.
Do this question again when the preceding layers are modified in the following way. The laminate is constituted of three layers. Layers 1 and 3 are double layers (thicknesses of 2 mm) of the previous mat reinforcement. Layer 2 is a double layer (thickness of 3.2 mm) of the previous unidirectional reinforcement. 20.1.3 Do the question 20.1.2 again in the case of a beam with one end clamped (the end x = 0) and the other end free (the end x = L).
20.2 Do Exercise 20.1 again in the case of a beam subjected to a transverse load varying linearly along the length of the beam: p(0) = 0 at the end x = 0 and p(L) = p0 at the end x = L.
Exercises
431
20.3 The five-layers symmetric laminate considered in Exercise 20.1 forms half the skins of a symmetric sandwich beam the isotropic core of which has a thickness of 30 mm and the mechanical characteristics:
Do Exercise 20.1 again in the case of this sandwich beam.
CHAPTER 21
Bending of Orthotropic Laminate Plates
21.1 INTRODUCTION
The analysis of the behaviour of plates constituted of laminate or sandwich materials has varying degrees of complexity. Cylindrical bending and beam bending (Chapters 19 and 20), reduced to one-dimensional analyses, are the easiest problems to study. In the study of plate bending, the most complex analysis is that of laminates made with an arbitrary stacking sequence, presenting in-plane-bending, in-planetwisting and bending-twisting couplings. The first simplification consists in the study of symmetric laminates for which there exists no in-plane-bending/twisting couplings: the terms Bij are zero (Bij = 0). An additional simplification occurs in the case where there exists no bending-twisting coupling: the terms D16 and D26 are zero (D16 = D26 = 0). The symmetric laminates (Bij = 0), for which there exists no bending-twisting coupling (D16 = D26 = 0), are referred as orthotropic laminates. This type of laminate is obtained either in the case of a single layer of orthotropic material or in the case of a symmetric laminate made of orthotropic layers the directions of which coincide with the laminate directions, such as the case of cross-ply laminates, for example (Chapter 15). In this chapter, we consider the analysis of the bending of plates made of orthotropic laminates.
21.2 SIMPLY SUPPORTED RECTANGULAR PLATES 21.2.1 General Expressions

We consider a rectangular plate subjected to a distributed transverse load: q = q(x, y) (Figure 21.1). In the case of an orthotropic laminate, Relations (16.7) to (16.9) reduce to:
21.2 Simply Supported Rectangular Plates
433
q ( x, y )
b
a y z
FIGURE 21.1. Rectangular plate subjected to a distributed load.
A11
2 u0 x 2
+ A66
2 u0 y 2
+ ( A12 + A66 )
2v0 = 0. xy
(21.1)
2 u0 2v0 2v0 + A66 2 + A22 2 = 0. ( A12 + A66 ) xy x y D11 4w 0 x 4 + 2 ( D12 + 2 D66 ) 4w 0 x 2 y 2 + D22 4w 0 y 4 = q.
(21.2) (21.3)
For a plate simply supported along its four edges, the boundary conditions are written as: edges x = 0 and x = a : edges y = 0 and y = b : w0 = 0, w0 = 0, Mx = 0, My = 0. (21.4) (21.5)
From the constitutive equation (14.29), the conditions on the bending moments along the edges are: 2w 0 2w 0 edges x = 0 and x = a : (21.6) M x = D11 D12 =0, x 2 y 2 edges y = 0 and y = b :
M y = D12 2w 0 x 2 D22 2w 0 y 2 = 0.
(21.7)
As the supports are simple, there are no conditions imposed on u0 and v0. In the general case, the transverse load can be expanded as a double Fourier series as: q ( x, y ) =
qmn sin m a sin n b ,

m =1 n =1
(21.8)
434
Chapter 21 Bending of Orthotropic Laminate Plates
where the coefficients qmn are given by:

qmn = 4 ab

x =0
b y =0
q ( x, y ) sin m
x y sin n d x d y . a b
(21.9)
The solutions of the plate bending problem can thus be investigated by expressing the transverse displacements in the form of a double series, satisfying the boundary conditions. For example: u0 ( x, y ) = v 0 ( x, y ) = w 0 ( x, y ) =
Amn cos m a cos n b ,

m =1 n =1
(21.10)
Bmn cos m a cos n b ,

m =1 n =1
(21.11)
Cmn sin m a sin n b .

m =1 n =1
(21.12)
These expressions for u0 and v0 substituted in Relations (21.1) and (21.2) imply that Amn = 0 and Bmn = 0. So, the in-plane displacements are zero: u0 = 0, v0 = 0. This is a general result in the case of laminates not having in-plane and flexural coupling. The expression of the coefficient Cmn is obtained by substituting Expression (21.12) for w0 into Relation (21.3) and by expressing the transverse load q(x, y) according to (21.8). We obtain: qmn
Cmn = D11
( )
m a
m + 2 ( D12 + 2 D66 ) a

( ) ( ) +D ( )
2
n b
22
n b
(21.13)
The deflection at point (x, y) is thus written as:

w 0 ( x, y ) =
a4
m =1 n =1
qmn x y sin m sin n , Dmn a b
(21.14)
where
Dmn = D11m 4 + 2 ( D12 + 2 D66 ) m 2 n 2 R 2 + D22 n 4 R 4 ,
(21.15)
on introducing the length-to-width ratio of the plate: R = a/b. The expressions for the moments are next obtained by substituting Equation (21.14) into the constitutive equation (14.29): Mx = a2
m =1 n =1
qmn ( m2 D11 + n2 R 2 D12 ) sin m x sin n y , Dmn a b
(21.16)

435
My =
a2
m =1 n =1
qmn ( m2 D12 + n2 R 2 D22 ) sin m x sin n y , (21.17) Dmn a b
M xy = 2
a2
RD66
mn Dmn mn
m =1 n =1
cos m
x y cos n . a b
(21.18)
The in-plane stresses are deduced from Expressions (14.20), hence:

k xx =
a2
2
a2
m =1 n =1

qmn k k ( m2Q11 ) sin m x sin n y , (21.19) + n 2 R 2Q12 Dmn a b qmn k k ( m2Q12 ) sin m x sin n y , (21.20) + n 2 R 2Q22 Dmn a b
k yy =
z a2
m =1 n =1 k RQ66 2
k xy
= 2
mn Dmn mn
m =1 n =1
cos m
x y cos n . a b
(21.21)
k k and yz in layer k can be determined next by The transverse shear stresses xz
substituting Equations (21.19) to (21.21) into Relations (8.20), and then integrating with respect to z. The constants of integration are determined by considering the continuity of the shear stresses between the layers and their vanishing on the upper (or lower) face. Expressions (21.14) to (21.18) show that, for a laminate where D22 = D11, the displacement field and the moment field are unchanged, R is changed into 1/R, that is, when the length and width are interchanged.
21.2.2 Case of Uniform Load

21.2.2.1 Analytical Expressions
In the case of a uniform load: q(x, y) = const = q0, Expression (21.9) leads to:
qmn = qmn 16q0 mn 2 = 0, , if m and n are odd, if m and n are even.
The two expressions can be regrouped in the following form:

q2 m1, 2 n 1 = 16q0
1 , ( 2m 1)( 2n 1)
m, n = 1, 2, . . . .
(21.22)
436
The deflection (21.14) at point of coordinates (x, y) of the plate is then written as:
w 0 ( x, y ) =
16a 4 q0
m =1 n =1

x y sin ( 2m 1) sin ( 2n 1) a b , ( 2m 1)( 2n 1) D2 m1, 2 n1 x y
(21.23)
or
w 0 ( x, y ) =
16a 4 q0
cmn sin ( 2m 1) a sin ( 2n 1) b ,

m =1 n =1
(21.24)
with
cmn = 1 . ( 2m 1) ( 2n 1) D2 m 1, 2 n 1
(21.25)
The in-plane stresses in layer k are deduced from Relation (14.20): 2w 0 2 x 2w 0 , 2 y 2 2 w 0 xy x y
xx Q11 Q12 = z Q Q22 yy 12 0 0 xy k with 2w 0 x

2
0 0 Q66 k
(21.26)
16a 2 q0
( 2m 1)2 cmn sin ( 2m 1) a sin ( 2n 1) b ,

m =1 n =1
2w 0 y
2
16a 2 R 2 q0
( 2n 1)2 cmn sin ( 2m 1) a sin ( 2n 1) b , ,

m =1 n =1
2w 0 32a 2 Rq0 2 = xy 4
( 2m 1) ( 2n 1) cmn cos ( 2m 1) a cos ( 2n 1) b .

m =1 n =1
The deflection is maximum at the centre of the plate: x = a/2, y = b/2. The numerators in Relation (21.23) are then:
sin ( 2m 1)
sin ( 2n 1)
= ( 1)
m+ n2
and the maximum deflection is given by:

w 0 max = 16a 4 q0
(21.27)
introducing the factor:

437
mn ,
m =1 n =1
mn = ( 1
)m + n + 2
cmn
( 1)m+ n+ 2 = . ( 2m 1)( 2n 1) D2 m1, 2 n 1
(21.28)
From Relation (21.26), the in-plane stresses at the plate centre are expressed as: xx Q11 Q12 = z Q Q22 yy 12 0 0 xy k with 0 16a 2 q0 0 4 Q66 k 2 R , 0 (21.29)
= =
Whence:
mn ,
m =1 n =1
mn = ( 2m 1)2 mn , mn = ( 2n 1)2 mn .
(21.30) (21.31)
mn ,
m =1 n =1
k xx k yy
= =
16a 2 q0
k k ( Q11 ) z, + R 2 Q12 k k ( Q12 ) z, + R 2 Q22
16a 2 q0
4
(21.32)
k = 0, xy
or still:
k k xx = Axx z, k k yy = Ayy z, k xy = 0,
(21.33)
with
k Axx = k Ayy
16a 2 q0
k k ( Q11 ), + R 2 Q12 k k ( Q12 ). + R 2 Q22
(21.34) (21.35)
16a 2 q0
4
21.2.2.2 Example
We consider the case of a rectangular plate with length a = 2.8 m and width b = 0.7 m, subjected to a uniform pressure of 500 Pa (Figure 21.2a). The plate is made of a symmetric laminate consisting of five layers arranged as in Figure
438
21.2b. Layers 1, 3 and 5 are layers reinforced with mats of weight per unit area equal to 450 g/m2, the characteristics of which were determined in (15.84):
EL = ET = 7.72 GPa,
LT = 0.33,
GLT = 2.91 GPa.
(21.36)
Layer 3 is a double layer. Layers 2 and 4 are double layers reinforced with a cloth of weight 500 g/m2, the characteristics of which were determined in (15.74):
EL = ET = 13.8 GPa,
LT = 0.12,
GLT = 1.87 GPa.
(21.37)
The reduced stiffness coefficients are deduced from Relations (11.52): for the mat layers 1, 3, 5:
Q11 = 8.66 GPa, Q22 = Q11 = 8.66 GPa, for the cloth layers 2, 4: Q11 = 14.00 GPa, Q22 = Q11 = 14.00 GPa, Q12 = 1.68 GPa, Q66 = 1.87 GPa. (21.39) Q12 = 2.86 GPa, Q66 = 2.91 GPa, (21.38)
The flexural stiffness coefficients Dij are derived from Relation (14.33). Thus:
D11 = 272.64 Nm, D22 = D11 = 272.64 Nm,
D12 = 64.834 Nm, D66 = 67.358 Nm.
(21.40)
From Equation (21.15), the coefficient D2m1, 2n1 is expressed as:

4 2 2 4 D2 m1, 2 n1 = 272, 64 ( 2m 1) + 6385, 64 ( 2m 1) ( 2n 1) + 69796, 68 ( 2n 1) ,
(21.41) The maximum deflection w0max is next deduced from Expression (21.27). The
q0 = 500 Pa
5 4
mat cloth cloth mat mat cloth cloth mat
1 mm 1.4 mm 2 mm 1.4 mm 1 mm
b = 0.7 m
2
a = 2.8 m
(a)
(b)
FIGURE 21.2. Rectangular plate subjected to a uniform pressure.
439
TABLE 21.1. Value of the coefficient as a function of m and n. m n
105 ( Pa 1 )
1.3080 1.0807 1.1302 1.1169 1.1213 1.1196 1.1203 1.1199 1.1200 1.1201 1.1201
1 2 3 4 5 6 7 8 10 15 20
1 2 3 4 5 6 7 8 10 15 20
values of the series as a function of m and n are reported in Table 21.1. This series converges rapidly, yielding:
= 1.1201 105 Pa 1 .
(21.42)
Whence the maximum deflection with a = 2.8 m and q0 = 500 Pa:

w 0 max = 5.728 mm .
(21.43)
The calculation of the stresses according to (21.33) needs the determination of series (21.30) and (21.31). Their numerical values are reported as functions of m and n in Table 21.2. We observe a slow convergence of the series . Finally, we have: = 1.1379 106 Pa 1 , = 1.0894 105 Pa 1. (21.44) The in-plane stresses in the mat layers, at the plate centre, are expressed as:
m m xx = Axx z, m m yy = Ayy z, m xy = 0,
(21.45)
with
m Axx = m Ayy =
16a 2 q0
m m ( Q11 ), + R 2 Q12 m m ( Q12 ), + R 2 Q22
16a 2 q0
4
(21.46)
m where the values of the parameters Qij are given in (21.38). Considering the
numerical values leads to:
440
TABLE 21.2. Values of the coefficients and as functions of m and n. m 1 2 3 4 5 6 7 8 10 15 30 60 100 150 n 1 2 3 4 5 6 7 8 10 15 30 60 100 150
106 ( Pa 1 )
13.080 6.904 5.383 1.1328 2.4327 3.5091 1.6454 0.7943 0.9604 1.1910 1.1313 1.1371 1.1378 1.1379
105 ( Pa 1 )
1.3080 1.0484 1.1000 1.0865 1.0906 1.0890 1.0897 1.0893 1.0894 1.0895 1.0894 1.0894 1.0894 1,0894
m Axx = 3.2721 108 Nm 1 ,
m Ayy = 9.744 108 Nm 1.
(21.47)
Similarly, the stresses in the cloth layers are expressed as:

c c xx = Axx z, c c yy = Ayy z , c xy = 0,
(21.48)
with
c Axx = c Ayy =
16a 2 q0
c c ( Q11 ), + R 2 Q12 c c ( Q12 ), + R 2 Q22
16a 2 q0
4
(21.49)
c are given in (21.39). The numerical where the values of the stiffness constants Qij
application leads to:

c Axx = 1.9883 108 Nm 1 , c Ayy = 1.5727 108 Nm 1.
(21.50)
The stress variation through the thickness of the plate and at its centre is reported in Figure 21.3. The transverse shear stresses are derived by substituting the in-plane stresses (21.26) in Relations (8.20), that is:
xx xy xz + + = 0, x y z
(21.51)
441
1.112 3
3.313 3
mat
0.477 0.785
mat
3.774 2.339
z ( mm )
cloth
0.327 0.199
cloth
0.974 1.573
mat
0 0 0.4 0.8 1.2 0 0 1 2
mat
3
xx ( MPa )
yy ( MPa )
FIGURE 21.3. Stress variations at the plate centre.
xy x
yy y
yz z
= 0.
(21.52)
The transverse shear stresses at point (x, y) of the plate are then written in the forms:
k k xz = Bxz ( x, y ) z 2 + const ,
(21.53)
k k yz = Byz ( x, y ) z 2 + const ,
(21.54)
k = m, c. The constants are derived by considering the continuity of the stresses between the layers and their vanishing on the lower and upper faces. We obtain: Layer 1 (mat)
1 m 2 ), iz = Biz ( x, y ) ( z 2 h0
i = x, y ,
h0 = 1 mm .
(21.55)
Layer 2 (cloth)
2 c 2 m 2 2 ) + Biz ), iz = Biz ( x, y ) ( z 2 h1 ( x, y ) ( h1 h0
i = x, y ,
h1 = 2.4 mm.
(21.56)
Layer 3 (mat)
3 m 2 2 2 c 2 2 ) + Biz ), iz = Biz ( x, y ) ( z 2 h2 + h1 h0 ( x, y ) ( h 2 h1
i = x, y ,
h2 = 1 mm.
(21.57)
442
The stress variation in layers couches 4 and 5 are symmetric to the variation in layers 2 and 1.
21.2.3 Case of a Load Distributed over a Rectangle

Another interesting case of loading is the one of a load P distributed over a rectangle (Figure 21.4), with centre (x0, y0) and sides c and d. In this case, the coefficients qmn (21.9) are expressed as:
qmn
4 = ab
x0 +
c 2 c 2
x = x0
y0 +
d 2 d 2
y = y0
q ( x, y ) sin m
(21.58)
In the case of a load uniformly distributed over the rectangle, the load density q(x, y) is constant. So: P q ( x, y ) = q0 = . (21.59) cd Expression (21.58) then leads to:
qmn = 16q0 mn 2 16 P mn cd
2
sin m
x0 m c y n d , sin sin n 0 sin a 2 a b 2 b x0 m c y n d . sin sin n 0 sin a 2 a b 2 b
(21.60)
or
qmn =
sin m
(21.61)
y a
c d b
y0
x0
FIGURE 21.4. Load distributed over a rectangle.
21.3 Rectangular Plates with Two Simply Supported Edges
443
The case of a concentrated load P located at point (x0, y0) is obtained by making c and d approach zero. In this case we obtain: qmn = 4P x y sin m 0 sin n 0 . ab a b (21.62)
The expressions for the deflection, the moments and the stresses at point (x, y) are next obtained by substituting the expressions for the coefficients qmn into Relations (21.14) to (21.21). For example, in the case of a load P concentrated at point (x0, y0), the deflection at point (x, y) is written as:
w 0 ( x, y ) =
4a 2 RP
m =1 n =1
1 x y x y sin m 0 sin n 0 sin m sin n . Dmn a b a b (21.63)
The deflection at the centre of the plate, using the results already introduced, is expressed as:
w 0 (a 2, b 2) =
4a 2 RP
m =1 n =1
( 1)m+ n + 2
D2 m1, 2 n1
x y sin ( 2m 1) 0 sin ( 2n 1) 0 . a b (21.64)
21.3 RECTANGULAR PLATES WITH TWO SIMPLY SUPPORTED EDGES 21.3.1 Case of an Arbitrary Load
In this subsection, we consider the case of a rectangular plate, subjected to a distributed transverse load q = q(x, y) and simply supported along two edges: the edge y = 0 and the edge y = b. The conditions along the edges x = 0 and x = a are not defined for the moment. The notation is identical to that of Figure 21.1. The load is expanded as a double Fourier series (21.8), and the solution for the deflecttion is searched in the form:
w 0 ( x, y ) =
n =1
y a4 n (x) sin n + 4 b
m =1 n =1
qmn x y sin m sin n , Dmn a b
(21.65)
where Dmn was defined in (21.15) and n(x) is a function to be determined. The second term is the solution found in (21.14). Expression (21.65) satisfies the conditions along the two simply supported edges. By substituting Expression (21.65) into the bending equation (21.3), and taking (21.8) into account, we obtain:
n =1
d 4 n n 2 2 d 2n n 4 4 y + + 2 D 2 D D n sin n = 0 . (21.66) ( ) D11 12 66 22 4 2 2 4 b b b dx dx
444
This expression is satisfied if n(x) is solution of the differential equation:

D11 d 4n d x4 2 ( D12 + 2 D66 ) n 2 2 d 2n b2 d x2 + D22 n 4 4 b4
n = 0 .
(21.67)
The solutions of this differential equation have the form:
n ( x) = exp n .
x b
(21.68)
The equation allowing us to find the coefficient is obtained by substituting (21.68) into (21.67). Whence: D11 4 2 ( D12 + 2 D66 ) 2 + D22 = 0 , which is the characteristic equation. The roots are of the form: 1 2 D12 + 2 D66 ( D12 + 2 D66 ) D11D22 . D11 (21.69)
2 =
(21.70)
The final form of n(x) depends on the nature of the roots .

1. Case of real and different roots
In the case where Equation (21.70) leads to real and different roots:
= r1 and = r2 ,
with r1, r2 > 0, the solutions of (21.67) are of the form: x b x b x b x b
(21.71)
n ( x) = An cosh n r1 + Bn sinh n r1 + Cn cosh n r2 + Dn sinh n r2 .

(21.72) The deflection is then written as:
w 0 ( x, y ) =
An cosh n r1 + Bn sinh n r1 + Cn cosh n r2 b b b x x

n =1 4
x (21.73)
x a qmn x y sin m sin n . + Dn sinh n r2 + 4 b m=1 Dmn a b
2. Case of real and equal roots
In the case where Equation (21.70) leads to equal real roots:
= r,
x b
r > 0,
(21.74)
the solutions of the differential equation (21.67) are written as:
n ( x) = ( An + Bn x ) cosh n r + ( Cn + Dn x ) sinh n r .
x b
(21.75)
445
The deflection in this case is given by:

w 0 ( x, y ) =
( An + Bn x ) cosh n r + ( Cn + Dn x ) sinh n r b b
x
n =1
x (21.76)
a4
m=1
qmn sin m Dmn
x y sin n . a b
3. Case of complex roots In the case of complex roots:
= r1 i r2
and
= r1 i r2 ,
(21.77)
with r1, r2 > 0, the function n(x) is written as: x x x n ( x) = An cos n r2 + Bn sin n r2 cosh n r1 b b b x x x + Cn cos n r2 + Dn sin n r2 sinh n r1 . b b b The deflection is then expressed as:
w 0 ( x, y ) = An cos n r2 + Bn sin n r2 cosh n r1 b b b x x
n =1
(21.78)
x x x + Cn cos n r2 + Dn sin n r2 sinh n r1 b b b a4 qmn x y + 4 sin m sin n . a b m=1 Dmn
(21.79)
4. Application In the three cases considered previously, the constants An, Bn, Cn and Dn are determined so as to satisfy the conditions imposed along the edges x = 0 and x = a (free edges, clamped edges, etc.). For example, in the case of clamped edges (16.30), the conditions to be satisfied for x = 0 and x = a are:
w 0 = 0,
w 0 = 0. x
(21.80)
We consider the case where the characteristic equation has real and different roots. In this case, by (21.73) we deduce: w 0 ( x, y ) = x

n =1
x x n r2 x n r1 An sinh n r1 + Bn cosh n r1 + Cn sinh n r2 b b b b b
x a3 q x y m mn cos m sin n . + Dn cosh n r2 + 3 b m=1 Dmn a b
(21.81)
446
The clamping conditions along the edges lead to: for x = 0

0 = An + Cn ,
(21.82)
n r1 n r2 a3 q 0= Bn + Dn + 3 m mn , b b m=1 Dmn
(21.83)
for x = a
0 = An cosh n r1R + Bn sinh n r1R + Cn cosh n r2 R + Dn sinh n r2 R
(21.84)
0=
n r1 ( An sinh n r1R + Bn cosh n r1R ) b n r2 a3 qmn x + + cos , m m ( Cn sinh n r2 R + Dn cosh n r2 R 3 m=1 Dmn b a
(21.85)
where R is the length-to-width ratio of the plate. Equations (21.82) to (21.85) allow to determine the constants An, Bn, Cn and Dn for n = 1, 2, etc.
21.3.2 Case of Uniform Loading

In the case of a uniform load: q(x, y) = q0, the solution can be achieved by expanding the load in a single Fourier series of the form: q ( x, y ) = with qn = Which leads to:
4q0 , n qn = 0, qn = if n is odd, if n is even.
qn sin n b ,
n =1
(21.86)
2 b
b 0
q( x, y ) sin n
y dy . b
(21.87)
Hence:
q ( x, y ) = 4q0
n =1,3,...
1 y sin n . n b
(21.88)
The second term in Expression (21.65) simplifies and the deflection may be written in the form: 4a 4 q0 1 y (21.89) sin n , w 0 ( x, y ) = n ( x) + 5 4 5 b R D22 n n =1,3,...
447
where the functions n(x) have been determined in (21.72), (21.75) and (21.78). We consider the case of clamped edges along x = 0 and x = a, and the case where the characteristic equation (21.69) has two real and different roots. In this case, we have:
w 0 ( x, y ) =
n =1,3,...
x x An cosh n r1 + Bn sinh n r1 b b
4
4a q x x y + Cn cosh n r2 + Dn sinh n r2 + 5 5 40 sin n . b b n R D22 b w 0 x x n r1 ( x, y ) = An sinh n r1 + Bn cosh n r1 b b x b n =1,3,...
(21.90)
n r2 x x y Cn sinh n r2 + Dn cosh n r2 sin n . b b b b
(21.91)
The clamping conditions are written as: for x = 0

An + Cn + 4a 4 q0 n5 5 R 4 D22 = 0,
(21.92) (21.93)
r1Bn + r2 Dn = 0 ,
for x = a An cosh n r1R + Bn sinh n r1R + Cn cosh n r2 R + Dn sinh n r2 R + 4a 4 q0 n R D22

5 5 4
= 0,
(21.94)
r1 An sinh n r1R + r1Bn cosh n r1R + r2Cn sinh n r2 R + r2 Dn cosh n r2 R = 0. (21.95)
Solving the system of Equations (21.92) to (21.95) leads to: An = 4r2 a 4 q0 n5 5 R 4 D22 H n 4r2 a 4 q0 n5 5 R 4 D22 H n
[ r1 ( cosh n r2 R 1)( cosh n r1R cosh n r2 R ) [ r2 ( cosh n r1R cosh n r2 R ) sinh n r2 R

(21.96)
( r2 sinh n r1R r1 sinh n r2 R ) sinh n r2 R ] ,
Bn =
( cosh n r2 R 1)( r1 sinh n r1R r2 sinh n r2 R )] ,
4a 4 q Cn = An + 5 5 4 0 , n R D22 r Dn = 1 Bn , r2
448
with
H n = r1r2 ( cosh n r1R cosh n r2 R )
( r2 sinh n r1R r1 sinh n r2 R )( r1 sinh n r1R r2 sinh n r2 R ) . In Expressions (21.96) for the coefficients An, Bn, Cn and Dn, the ratio R is the length-to-width ratio of the plate, and n takes odd values: n = 1, 3, 5, etc.
21.4 RECTANGULAR PLATES WITH VARIOUS BOUNDARY CONDITIONS ALONG THE EDGES
In the two cases studied in the preceding sections (simply supported plates), the bending problem has been solved by expressing the exact solutions of the fundamental equations satisfying the boundary conditions along simply supported edges. In the case of a rectangular plate with various support conditions, combinations of boundary conditions such as simple supports, clamped edges, free edges, etc., the exact analytical solutions of the bending problem of the plates cannot be derived. Therefore approximation methods have to be used. In this section, we investigate approximate solutions using the Ritz method (Section 8.4). In the case of orthotropic plates, the strain energy is given by Relation (16.38): 1 Ud = 2

x =0
b y =0
2 2 2 2 2 2 D11 w 0 + 2 D12 w 0 w 0 + D22 w 0 x 2 y 2 x 2 y 2 (21.97) 2 2 w0 + 4 D66 xy d x d y + C.
The work of the distributed transverse load: q = q(x, y) is written (Relation (16.45)) as:
Wf =

x =0
b y =0
q ( x, y ) w 0 ( x, y ) d x d y .
(21.98)
The approximate solution is searched as a double series expressed in the separated variables form:
w 0 ( x, y ) =
Amn X m ( x)Yn ( y) .
m =1 n =1
(21.99)
The functions Xm(x) and Yn(y) have to form functional bases (Subsection 8.4.2), such as polynomials, trigonometric functions, hyperbolic functions, etc., and are
21.4 Rectangular Plates with Various Boundary Conditions along the Edges
449
chosen so as to satisfy the boundary conditions. The coefficients Amn are next determined by the stationarity conditions (8.62), which here are written:
U =0 Amn
or
d f U W , = Amn Amn
(21.100)
d and W f are the strain energy and the work of the external loads, where U obtained by substituting the approximate expression (21.99) of the deflection into Expressions (21.97) and (21.98), respectively. The calculation of the approximate strain energy requires to express the terms:
2w 0 2 , x For example:
2w 0 2w 0 x 2 y 2 2w 0 x 2
M
2w 0 2 , y Amn d2 X m d x2 Yn .
2w 0 . xy (21.101)
m=1 n =1 M N
Whence
M N 2w 0 2 = x m=1 n =1 2
Amn Aij
i =1 j =1 M 2
d2 X m d2 X i d x2 d x2
YnY j ,
(21.102)
and 1 2w 0 = 2 Amn x 2 The integration of this term yields: 1 2 Amn
i =1 j =1 M N
Aij
YnY j .
(21.103)

x =0
b y =0
2w 0 2 dx dy = x
i =1 j =1
Aij
a 0
dx
b 0
YnY j d y . (21.104)
In order to express these integrals, it is useful to introduce the reduced variables: u= x a and
v=
y . b
(21.105)
Expression (21.104) may then be written in the form: 1 2 Amn

x =0
M
b y =0
N
2w 0 2 dx dy = x
Aij b a3
(21.106)
du
i =1 j =1
1 0
d2 X m d2 X i d u2 d u2
1 0
YnY j d v.
The integrals are then dimensionless.
450
Proceeding in the same way for the other terms, the left-hand side of Expression (21.100) can be put into the practical form:
d U 1 = Amn Ra 2
22 00 20 02 02 20 J nj + D12 ( I mi J nj + I mi J nj ) {D11I mi i =1 j =1 11 11 2 00 22 4 J nj R + D22 I mi J nj R Aij , + 4 D66 I mi M N
(21.107)
where R is the length-to-width ratio of the plate (R = a/b) and where we have introduced the dimensionless integrals:
pq = I mi
1 0
1 0
d p X m dq X i d u p d uq
s d rYn d Y j
du
m, i = 1, 2, . . . , M , pq = 00, 02, 11, 20, 22.

n, j = 1, 2, . . . , N , rs = 00, 02, 11, 20, 22.
(21.108)
rs J nj
dv
dv
dv
(21.109)
The approximate expression for the load of the transverse load is, by Expressions (21.98) and (21.99), written as:
f = W
m=1 n =1
Amn

x =0 M
b y =0
X m ( x) Yn ( y ) q( x, y ) d x d y ,
(21.110)
or
f = ab W
00 (q) , Amn I mn m=1 n =1 N
(21.111)
on introducing the integral:

00 I mn (q) =

u =0
1 v =0
X m (u ) Yn (v ) q(u, v ) d u d v .
(21.112)
From this it results:

f W a 2 00 = I mn (q) , Amn R
(21.113)
Finally, Expression (21.100) leads to the system of equations:

22 00 20 02 02 20 11 11 2 J nj + D12 ( I mi J nj + I mi J nj ) + 4 D66 I mi J nj R {D11I mi i =1 j =1 00 22 4 00 + D22 I mi J nj R Aij = a 4 I mn (q), M N
(21.114)
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
The system of equations can also be rewritten in a reduced form as:
21.5 Clamped Rectangular Plates
451
2200 2002 0220 1111 2 + D12 ( Cminj + Cminj ) + 4D66Cminj {D11Cminj R i =1 j =1 0022 4 00 + D22Cminj R Aij = a 4 I mn (q),
(21.115)
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N ,
on expressing the product of integrals (21.108) and (21.109) in the form:

pqrs pq rs Cminj = I mi J nj =
1 0
d p X m dq X i d u p d uq
du
1 0
s d r Yn d Y j
dvr dvs
dv .
(21.116)
In the system of Equations (21.114) or (21.115), the integrals are calculated, considering the boundary conditions and the transverse load q(x, y) applied. Solving the system next allows us to find the coefficients Aij and to deduce the transverse displacement expressed at each point (x, y) using Expression (21.99). In the case of a uniform load: q(x, y) = q0, the integral on the right-hand side of the system (21.114) is given by:
00 0 0 I mn (q0 ) = q0 I m In ,
(21.117)
on introducing the integrals:

0 Im =
1 0
X m d u,
0 Jn =
1 0
Yn d v.
(21.118)
In this case, the system of Equations (21.115) is written as:

2200 2002 0220 1111 2 + D12 ( Cminj + Cminj ) + 4D66Cminj {D11Cminj R i =1 j =1 0022 4 0 0 + D22Cminj R Aij = a 4 I m I n q0 , M N
(21.119)
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
21.5 CLAMPED RECTANGULAR PLATES 21.5.1 Introduction

As an application of the general formulation developed in the previous section, we consider hereafter the case of a rectangular plate clamped along the four edges and subjected to a uniform transverse load q0 (Figure 21.5). As the plate is clamped along the four edges, the boundary conditions are:
452
for the edges x = 0 and x = a :

w 0 = 0,
w 0 = 0, x w 0 = 0. y
(21.120)
for the edges y = 0 and y = b :

w 0 = 0,
(21.121)
Hereafter, we consider the case of two types of functions Xm(x) and Yn(y) satisfying these conditions.
21.5.2 Solution Approximated by Polynomial Functions

For the functions Xm(x) and Yn(y), we choose the polynomial functions of the form:
X m ( x) = x2 x x 1 a2 a a
2 2 m 1
= u 2 ( u 1) u m1 ,
2
y2 y y Yn ( y ) = 2 1 b b b
n 1
(21.122)
=v
2(
v 1 v
)2 n1
These functions satisfy the boundary conditions (21.120) and (21.121). The identity of the functions (21.122) of the reduced variables u and v leads to the following relations between the integrals defined in (21.108), (21.109) and (21.118): rs rs 0 0 J nj = I nj , Jn = In . (21.123)
pq The integrals I mi can be expressed analytically, then calculated numerically for
x q0 b
a y z
FIGURE 21.5. Clamped rectangular plate.
453
the various values of m and i. They can also be calculated directly using a software package for scientific and engineering applications that incorporate numerical processing. Tables A.1 to A.4 of Appendix A give the numerical values 0 00 02 22 , I mi , I mi and I mi for m and i varying from 1 to 8. of respectively the integrals I m
11 02 and I mi The integrals I mi are the opposite of each other: 11 02 I mi = I mi .
(21.124)
In the case where M = N = 1, the approximation (21.122) is written:

x2 x 2 X1 ( x) = 2 1 = u 2 ( u 1) , a a y2 y 2 Y1 ( y ) = 2 1 = v 2 (v 1) , b b
2 2
(21.125)
and the system (21.119) reduces to a single equation: 22 00 20 2 11 2 2 00 22 4 4 0 2 D11I11 I11 + 2 D12 ( I11 ) + 2 D66 ( I11 ) R + D22 I11 I11 R A11 = a ( I1 ) q0 , (21.126) with
00 I11 =
1 2 2 4 1 11 20 22 0 , I11 = , I11 = , I11 = , I1 = . 630 105 105 5 30

6,125a 4 q0
(21.127)
So we obtain:
A11 = 7 D11 + 4 ( D12 + 2 D66 ) R 2 + 7 D22 R 4
2
(21.128)
Hence the expression for the deflection: x2 x y 2 y w 0 ( x, y ) = 1 1 . (21.129) 7 D11 + 4 ( D12 + 2 D66 ) R 2 + 7 D22 R 4 a 2 a b 2 b The deflection is maximum at the centre of the plate (x/a = y/b = by:
w 0 max = 0, 00342
D11 + 0,571( D12 + 2 D66 ) R 2 + D22 R 4 a 4 q0 .
1 2
6,125a 4 q0
) and is given (21.130)
21.5.3 Solution Approximated by Beam Functions

Another approximation, developed by Young [26], for obtaining the natural frequencies of isotropic rectangular plates, and adapted by Whitney [27] to the analysis of the bending of laminate plates, consists in expressing the functions Xm(x) and Yn(y) of the approximate solution in the form:
454
X m ( x) = cos m Yn ( y ) = cos n
x x x x cosh m m sin m sinh m , a a a a y y y y cosh n n sin n sinh n . b b b b
(21.131) (21.132)
These functions constitute the basis functions used to express the natural bending modes of beams clamped at both ends of the beam ([26] and Subsection 24.3.3). These functions satisfy the conditions along the edges x = 0 and y = 0: dXm X m x =0 = 0, = 0, d x x =0 (21.133) d Yn Yn y =0 = 0, = 0, d y y =0 It remains to satisfy the conditions along the edges x = a and y = b: Xm Yn
x=a
= 0, = 0,
y =b
dXm dx d Yn dy
x=a
= 0, (21.134) = 0.
y =b
For example, conditions (21.134) are written as: cos m cosh m m ( sin m sinh m ) = 0, sin m sinh m m ( cos m cosh m ) = 0. A solution different from m = 0 is obtained when: cos m cosh m sin m + sinh m = . sin m sinh m cos m cosh m Thus: (21.135)
( cos m cosh m )2 = sin 2 m sinh 2 m .

Taking into account the equalities:
cos 2 m + sin 2 m = 1, cosh 2 m sinh 2 m = 1,
the previous equation reduces to:

cos m cosh m = 1 .
From this it results that, to verify the conditions along the edges x = a and y = b, the coefficients m and n have to be the roots of the equation:
cos i cosh i = 1, i = m, n.
(21.136)
The coefficients m and n are next deduced from the relation:
455
TABLE 21.3. Values of the constants i and i of the beam function in the case of two clamped ends of the beam.
i 1 2 3 4 5 6 7 8
i
4.7300408 7.8532046 10.9956078 14.1371655 17.2787596 20.4203522 23.5619449 26.7035376
i
0.98250222 1.00077731 0.99996645 1.00000145 0.99999994 1.00000000 1.00000000 1.00000000
i =
cos i cosh i . sin i sinh i
(21.137)
The values of i and the corresponding values of i are reported in Table 21.3 for i varying from 1 to 8. It should be noted that an approximate solution can be considered in the case where i is large enough. In fact, in this case: 1 cosh i exp i , 2 and Equation (21.135) is written as: cos i 2 0. exp i
The solutions for this latter equation are:
i = ( 2i + 1) .
2
(21.138)
The approximate solutions (21.138) are compared in Table 21.4 with the exact solutions of Equation (21.136). As the functions (21.131) and (21.132) are entirely determined, it is possible to evaluate the various integrals introduced in the system of Equations (21.115) or (21.119).
TABLE 21.4. Exact and approximate values of i.
1
Solution of Equation (21.119) Approximate solution (21.121) 4.730 4.712
2
7.853 7.854
3
10.996 10.996
4
14.137 14.137
5
17.279 17.279
456
The identity of the functions (21.131) and (21.132) of the reduced variables u and v leads to the relations expressed in (21.123). Next, we obtain the result:
00 = I mi
1 0
1 if i = m, X m Xi d u = 0 if i m.
(21.139)
The functions Xm and Xi are orthonormal in the integration operation under consideration. Similarly, we have:
22 I mi
1 0
d2 X m d2 X i du
2
du
4 du = 0
if i = m, if i m.
(21.140)
The second derivative functions are orthogonal. The values of the integrals are then deduced from the values of given in Table 21.3. They are reported in Table B.1 in Appendix B for i and m varying from 1 to 8. Furthermore, we have the relation:
Hence the relations:
1 0
d X m d Xi du = du du
1 0
Xm
d2 X i d u2
du .
(21.141)
20 02 I mi = I mi ,
11 02 I mi = I mi .
(21.142)
02 The numerical values of the integrals I mi are reported in Table B.2 in Appendix B. 0 are reported in Table B.3. Lastly, the values of the integrals I m
In the case where M = N = 1, 1 = 4.730 and 1= 0.9825. The approximation (21.99) then reduces to:
x x x x X1 ( x) = cos 4.73 cosh 4.73 0.9825 sin 4.73 sinh 4.73 , a a a a (21.143) y y y Y1 ( y ) = cos 4.73 cosh 4.73 0.9825 sin 4.73 sinh 4.73 . b b b The system (21.119) reduces to Equation (21.126), with:
00 11 20 11 I11 = 1, I11 = 12.30262, I11 = I11 = 12.30262, 22 0 I11 = 500.564, I1 = 0.8309.
Hence:
A11 = D11 + 0.6047 ( D12 + 2 D66 ) R 2 + D22 R 4 0.001379a 4 q0
(21.144)
457
Thus, the expression of the deflection is:

w 0 ( x, y ) = A11 X1 ( x)Y1 ( y ) ,
(21.145)
where the functions X1(x) and Y1(y) are expressed in (21.143). The deflection is maximum at the centre of the plate and is obtained by introducing X1(a/2) and Y1(b/2) in the preceding equation, which yields:
w 0 max = 0.00348
D11 + 0, 6047 ( D12 + 2 D66 ) R 2 + D22 R 4
a 4 q0
(21.146)
21.5.4 Comparison between the Approximate Solutions

The comparison between Expressions (21.130) and (21.146) for the maximum deflection shows a similarity of the results obtained in the case of the approximation with one term (M = N = 1). This comparison can be specified by considering the laminate of Example 21.2.2.2, for which the flexural coefficients Dij are given in (21.40): D11 = 272.64 Nm, D22 = D11 = 272.64 Nm, D12 = 64.834 Nm, D66 = 67.358 Nm.
For a length-to-width ratio equal to 4 (R = 4), the approximation (21.130) based on the polynomial functions leads to:
w 0 max
a q0
w 0 max
= 6.714 107 Nm 1 ,
(21.147)
whereas the approximation (21.146) by the beam functions leads to:

a q0
4
= 6.797 107 Nm 1 .
(21.148)
This results in a difference of 1.2 % between the two approximations. By using several terms in the approximation, the search for the approximate solution requires to establish the system of equations (21.119), then to solve it to obtain the coefficients Amn. The deflection is next given by Relation (21.99) and the maximum deflection at the centre of the plate is written as:
w 0 max =
Amn X m ( a 2) Yn (b 2) .
m =1 n =1
(21.149)
Table 21.5 shows the results obtained for M = N varying from 1 to 11, using an approximation either with polynomial or with beam functions and considering the previous values for the coefficients Dij . These results show, in the case of the approximation by polynomials, a quite rapid convergence of the approximate solution to the exact solution, which is reached in practice for M = N = 5.
458
TABLE 21.5. Convergence of the approximate solutions when M = N increases. M=N 1 3 5 7 9 11
w 0 max a 4 q0 (107 N 1m 1 )
Polynomials 6.714 5.911 5.972 5.973 5.973 5.973 Beam Functions 6.797 5.825 6.006 5.962 5.977 5,971
Thus:
w 0 max
a q0
= 5.973 107 Nm 1 .
(21.150)
The convergence is slower in the case of the approximation by the beam functions. In addition, the difference between the approximation by one term and the solution obtained with a large number of terms is 12.4 % in the case of approximation by polynomials and 13.8 % in the other case. If the approximations by polynomials or by beam functions lead to a similar convergence of the approximate solution, the orthogonality properties (21.139) and (21.140) of the beam functions introduce numerous zeros into the nondiagonal terms of the system of equations (21.119). Resolving this system is then greatly simplified. As the deflection is obtained, the bending moments can be determined from Expression (14.29). For example:
M x = D11 2w 0 x 2 D12 2w 0 y 2
(21.151)
The approximation (21.129) by polynomials leads to: Mx = 1, 75R 4 q0 D11 ( 6 x 2 6ax + a 2 ) ( y 2 by ) + D12 ( x 2 ax ) ( 6 y 2 6by + b 2 )
2 2
a4
D11 + 0,571( D12 + 2 D66 ) R 2 + D22 R 4 (21.152)
The moment at the middle of the edge x = 0 is written as:

M x (0, b 2) = 0.109a 2 q0 D11 D11 + 0,571( D12 + 2 D66 ) R 2 + D22 R 4
. (21.153)
Which, with the previous numerical values, leads to:

M x (0, b 2) = 5,857 103 a 2 q0 .
(21.154)
In the same way, the approximation (21.145) by beam functions leads to:
459
M x (0, b 2) = 5.221 103 a 2 q0 .
(21.155)
Using a large number of terms M = N = 11 in the series (21.99), we obtain, in the case of a polynomial approximation, the value:
M x (0, b 2) = 14.190 103 a 2 q0 .
(21.156)
We observe that the one-term approximation (M = N = 1) leads for the moment at the middle of the edge x = 0 to some very important differences from the exact value: 59% for the polynomial approximation and 63% in the other case. Here we recover a result analogous to that observed in the case of rectangular plate simply supported (Section 21.2.2.2), where the convergence is much slower in the estimate of the stresses. To illustrate more completely the approximations by polynomials and beam functions, we have reported in Tables 21.6 to 21.8 the results obtained in the case of the bending of an isotropic plate (Table 21.6), a balanced orthotropic plate (Table 21.7), and a unidirectional plate (Table 21.8). Different values of the length-to-width ratio of the plate have been considered.
TABLE 21.6. Maxima of deflections and moments in the case of an isotropic plate ( D11 = D22 = D12 + 2 D66 = D) , for length-to-width ratios equal to 1 and 2.
R=1 Polynomials M = N w 0 max D a 4 q0 (103 )

1 3 5 7 9 11 1.3292 1.2645 1.2653 1.2653 1.2653 1.2653 0.04253 0.05116 0.05128 0.05139 0.05133 0.05132
Beam Functions
w 0 max D a 4 q0
M x a 2 q0
M x a 2 q0
0.03763 0.04526 0.04839 0.04952 0.05011 0.05049
(103 )
1.3354 1.2526 1.2671 1.2645 1.2655 1.2652
R=2
Polynomials
M = N w 0 max D a 4 q0
Beam Functions
w 0 max D a 4 q0
M x a 2 q0
0.00567 0.01262 0.01420 0.01425 0.01432 0.01425
M x a 2 q0
0.00505 0.00930 0.01143 0.01244 0.01301 0.01335
(103 )
1 3 5 7 9 11 1.7723 1.5635 1.5835 1.5831 1.5832 1.5831
(103 )
1.7912 1.5415 1.5924 1.5797 1.5843 1.5825
460
TABLE 21.7. Maxima of deflections and moments in the case of a balanced orthotropic square plate ( D11 = D22 = 272.64 Nm, D12 = 64.834 Nm, D22 = 67.358 Nm).
Polynomials M=N 1 3 5 7 9 11
w 0 max D a q0
(106 N 1m 1 )
4
Beam Functions
2
M x a q0
w 0 max D a 4 q0
(106 N 1m 1 )
M x a 4 q0
5.1842 4.9542 4.9539 4.9540 4.9540 4.9540
0.04523 0.05281 0.05274 0.02810 0.05279 0.05275
5.2232 4.9130 4.9603 4.9514 4.9547 4.9536
0.04012 0.04730 0.05019 0.05117 0.05171 0.05203
TABLE 21.8. Maxima of deflections and moments in the case of a unidirectional plate ( D11 = D22 = 272.64 Nm, D12 = 64.834 Nm, D22 = 67.358 Nm), for length-to-width ratios equal to 1/2, 1 and 2.
R = 1/2 Polynomials M=N 1 3 5 7 9 11 R=1 Polynomials M=N 1 3 5 7 9 11 R=2 Polynomials M=N 1 3 5 7 9 11
w 0 max D a q0 (106 N 1m 1 )
4
Beam Functions
4
w 0 max D a q0 (105 N 1m 1 )
M x a q0
w 0 max D a 4 q0 (105 N 1m 1 )
M x a 4 q0
3.2519 2.4950 2.6412 2.6334 2.6320 2.6319
0.10406 0.07830 0.08498 0.08401 0.08408 0.08403
3.3023 2.4499 2.6862 2.6138 2.6396 2.6287
0.09304 0.07457 0.08499 0.08222 0.08397 0.08346
Beam Functions
4
w 0 max D a q0 (105 N 1m 1 )
M x a q0
w 0 max D a 4 q0 (105 N 1m 1 )
M x a 4 q0
2.5238 2.3585 2.3611 2.3613 2.3614 2.3613
0.08076 0.07872 0.07907 0.07916 0.07893 0.07907
2.5515 2.3339 2.3664 2.3595 2.3619 2.3610
0.07189 0.07400 0.07739 0.07782 0.07834 0.07854
Beam Functions
4
M x a q0
w 0 max D a 4 q0 (106 N 1m 1 )
M x a 4 q0
7.0954 6.7149 6.7163 6.7167 6.7166 6.7167
0.02270 0.03204 0.03212 0.03210 0.03220 0.03211
7.1673 6.6533 6.7277 6.7127 6.7179 6.7160
0.02019 0.02709 0.02959 0.03055 0.03107 0.03136
21.6 Simply Supported Sandwich Plates
461
21.6 SIMPLY SUPPORTED SANDWICH PLATES

We now consider the case of a rectangular sandwich plate, having two identical skins constituted of an orthotropic laminate the material directions of which coincide with the reference directions of the plate (D16 = D26 = 0) and of a core the material directions of which are also the same as the plate directions. This plate is subjected to the transverse load: q( x, y ) = q0 sin x y sin . a b (21.157)
The study of an arbitrary loading is deduced from this loading by decomposition into a double Fourier series according to (21.8). Because the sandwich material is symmetric, we have (18.30):
Bij = 0,
Cij = 0.
(21.158)
The fundamental bending relations are given by relations (17.27) to (17.29), (18.31) and (18.32), the coefficients Aij, Dij and Fij being defined by Expressions (18.22) to (18.30), associated with (18.14) and (18.15). Relations (17.27) and (17.28) imply for a symmetric sandwich plate:
u0 = 0,
v0 = 0,
(21.159)
and the fundamental bending relations reduce to:
y 2w 0 2w 0 hG13 x + + hG + + q = 0 , (21.160) 23 x x 2 y 2 y D11 2 x x

2
+ D66
2 x y
2
+ ( D12 + D66 )
2
2 y
2
+ w 0 = 0 , (21.161) hG13 x xy x
( D12 + D66 )
y y 2 x + w 0 = 0 . (21.162) D hG + D66 + 22 23 y y xy x 2 x 2
The problem is thus reduced to the determination of the three functions: w0, x and y. In the case of a plate simply supported along its four edges, the boundary conditions are written as: along the edges x = 0 and x = a :
w 0 = 0,
y = 0,
M x = D11
y x + D12 = 0, x y y x + D22 = 0. x y
(21.163)
along the edges y = 0 and y = b :

w 0 = 0,
x = 0,
M y = D12
(21.164)
462
These conditions are satisfied by functions of the form:
x y , a b x y y = B sin cos , a b x y w 0 = C sin sin . a b
x = A cos sin
(21.165)
Substituting Expressions (21.165) into Equations (21.160) to (21.162), we obtain:
h h G13 G23 A G23 B 2 h 2 + 2 C + q0 = 0, a b a b 2 2 2 h B + G13 C = 0, D11 2 + D66 2 + hG13 A + ( D12 + D66 ) ab a a b 2 2 2 h ( D12 + D66 ) A + D66 2 + D22 2 + hG23 B + G23 C = 0. ab b a b G13
These equations can be rewritten in the form:
(21.166)
a1 A + a2 B + a3C = 0, a2 A + a4 B + a5C = 0, a3 A + a5 B + a6C =

on setting:
a1 = D11 + D66 R 2 + G13 a2 = ( D12 + D66 ) R, a3 = G13 ha ha 2
(21.167)
a2
q0 ,
,
2
a4 = D66 + D22 R + G23 a5 = G23 ha
ha 2
(21.168)
,
a6 = h ( G13 + G23 R 2 ) ,
R,
where R is the length-to-width ratio of the plate. Solving the system of equations (21.167) leads to:
A= a 2 q0
2D
( a2 a5 a3a4 ) ,
(21.169)
21.6 Simply Supported Sandwich Plates
463
B=
a 2 q0
2D
2
( a2 a3 a1a5 ) ,
2 ( a1a4 a2 ),
(21.170) (21.171)
C=
a 2 q0
with
2 2 2 D = a1a4 a6 + 2a2 a3a5 a1a5 a4 a3 a6 a2 .
(21.172)
The functions (21.165) with the coefficients given by Relations (21.169) to (21.171) constitute the solutions of the bending problem. As an example, we consider the case of a sandwich plate constituted of two skins each made of a single layer of an orthotropic material (a unidirectional or a cloth reinforced composite). The layers are characterized by their moduli EL, ET, LT and GLT. The calculation of the flexural coefficients Dij, expressed by Relations (18.30) and (18.26), leads to:
D11 =
D22
D22
D66
or
hh1 ( h + h1 ) EL , 2 ET 2 1 LT EL hh ( h + h1 ) ET ET = 1 = D11, E 2 ET L 2 1 LT EL hh ( h + h1 ) LT ET = 1 = LT D22 , 2 ET 2 1 LT EL 1 = hh1 ( h + h1 ) GLT , 2 D22 = h3 ET , D66 = h3GLT ,
(21.173)
D11 = h3 EL , D12 = h3 LT ET ,
with
(21.174)
h1 ( h + h1 ) h
2
1
2 ET 2 1 LT E L
(21.175)
On substituting Expressions (21.174) of Dij into Relations (21.168), the coefficients ai are obtained as:
a2 a1 = 1 + 1 2 ET h3 , h a2 a4 = 4 + 4 2 ET h3 , h a2 = 2 ET h3 , a5 = 5 ET ha, a3 = 3 ET ha,
(21.176)
a6 = 6 ET h,
464
where we have introduced: E G 1 G 1 = L + R 2 LT , 1 = 2 13 , ET ET ET 1 G13 GLT 2 = LT + E R, 3 = E = 1, T T (21.177) 1 G23 GLT 2 , 4 = 4 = 2 + R , ET ET 1 G23 G G R = 4 R, 5 = 6 = 13 + 23 R 2 = 2 ( 1 + 4 R 2 ) . ET ET ET Substituting Expressions (21.176) into Relation (21.172), we obtain:
a2 a4 3 7 D = ET h d1 + d 2 2 + d3 4 , h h
on setting:
2 d1 = 1 4 6 6 2 , 2 2 d 2 = 2 2 3 5 + 6 (1 4 + 4 1 ) 1 5 4 3 , 2 2 d3 = 6 1 4 1 5 4 3 .
(21.178)
(21.179)
Similarly, we can write:

2 a1a4 a2
2 6 ET h c1 + c2
a2
a4 + c3 4 , h2 h
c3 = 1 4 .
(21.180)
with
2 c1 = 1 4 2 ,
c2 = 1 4 + 4 1 ,
(21.181)
It follows that the maximum deflection (the deflection at the centre of the plate) may be expressed as:
w 0 max =
() () E h d +d ( ) ()
a 2 q0
2 T
c1 + c2
1
a h a h
a h 2 a + d3 h + c3
4 4
(21.182)
The variation of the maximum deflection as a function of the ratio a/h is reported in Figure 21.6 for a square plate, in the case where the characteristics of the skins are: EL GLT = 4.5, = 0.45, LT = 0.3, (21.183) ET ET the shear moduli of the core are given by:
G13 G23 = = 0.04, ET ET
(21.184)
Exercises
465
0.10
0.08
Deflection w0max ET h3 / q0 a4
0.06
0.04
0.02
10
20
30
40
50
Length-to-thickness ratio a / h
FIGURE 21.6. Maximum deflection for a sandwich plate subjected to a transverse load.
and the thickness of the skins and sandwich are:
h1 = 0.1,
h ht = . 8
(21.185)
EXERCISES
21.1 The simply supported plate considered in Example 21.2.2.2 is subjected to a transverse load P = 1 kN, concentrated at point (x0 = a/4, y0 = b/4). Derive the deflection at point (x, y), at point (x0, y0), then at the centre of the plate. Derive the stresses in each layer at point (x, y), at point (x0, y0), then at the centre of the plate. 21.2 In the case of an orthotropic rectangular plate clamped along its four edges, derive the expressions for the integrals introduced in the system (21.114), when the polynomial functions (21.122) are used. Calculate them for m varying from 1 to 8.
466
21.3 Form the system (21.119) for M = N = 3 in the case of a clamped rectangular plate subjected to a uniform load, using the polynomial functions (21.122), then the beam functions (21.131) and (21.132). Take account of the properties (21.139) to (21.142) in the case of the beam functions. Compare the two systems obtained. 21.4 We consider the system of equations established in Exercise 21.3, in the case of the beam functions. Derive the system for a plate of length twice the width, and with the characteristics:
D12 = 0.08 D11, D22 = 0.50 D11, D66 = 0.12 D11.
Calculate the coefficients Aij (referred to a4q0/D11). Next derive the deflection at point (x, y), then at the centre of the plate. Compare this with the case of a plate simply supported along all edges.
21.5 We consider a rectangular plate clamped along two adjacent edges and free along the others. In order to solve the bending problem of the plate subjected to a uniform transverse load, we consider the beam functions (24.147) and (24.148), introduced in Chapter 24. Derive the integrals introduced in the system (21.119) for M and N varying from 1 to 3. Do Exercise 21.4 again in the present case.
CHAPTER 22
Bending of Plates Made of Symmetric, Cross-Ply, or Angle-Ply Laminates
22.1 SYMMETRIC LAMINATE PLATES 22.1.1 General Expressions

We consider a symmetric laminate plate subjected to a distributed load: q = q(x, y) (Figure 21.1). In the case of a symmetric laminate, the terms Bij are zero, hence the absence of in-plane flexural coupling. In contrast to orthotropic laminates, the coefficients D16 and D26 are not zero, so it results that the symmetric laminates present a bending-twisting coupling. In the case of plates of symmetric laminates, the bending equation (u0 = v0 = 0) is given by Equation (16.9):
D11 4w 0 x 4 + 4 D16 4w 0 x3y + 2 ( D12 + 2 D66 ) 4w 0 xy 3 4w 0 x 2y 2 4w 0 y 4
(22.1)
= q.
+ 4 D26
+ D22
This equation differs from the bending equation (21.3) for orthotropic laminates by the inclusion of the terms D16 and D26, introducing odd derivatives of w0. It results that, contrary to orthotropic laminates, the bending equation (22.1) cannot be solved in the case of simple supports by expanding the load as a double Fourier series (21.8), and solving w0 in the form (21.12). Because of these difficulties, it is necessary to search for approximate solutions by using the Ritz method (Section 8.4), following the same procedure as that used in Section 21.4. In the case of symmetric laminates, the strain energy is given by Expression (16.37): 1 Ud = 2
2 2 2w 0 2w 0 2w 0 2w 0 D11 + 2 D12 + D22 2 2 2 2 x x y x =0 y =0 y (22.2) 2 2 2 2 2 w0 w0 w0 w0 + 4 D16 + D26 + 4 D66 d x d y + C. 2 2 xy x y xy a b
468
Chapter 22 Bending of Plates Made of Symmetric, Cross-Ply, or Angle-Ply Laminates
As in (21.99), the approximate solution is expanded in a double series in separate variables: w 0 ( x, y ) =
Amn X m ( x)Yn ( y) .
m =1 n =1
(22.3)
The functions Xm(x) and Yn(y) are chosen so as to satisfy the boundary conditions, and the coefficients Amn are determined by the stationarity conditions (21.100). Expression (22.2) of the strain energy differs from Expression (21.97) by the introduction of the terms in D16 and D26. Proceeding as in Section 21.4, the lefthand side of Expression (21.100) is obtained by completing Expression (21.107) with the terms in D16 and D26. Whence:
d U 1 = Amn Ra 2
22 00 20 02 02 20 11 11 2 J nj + D12 ( I mi J nj + I mi J nj ) + 4 D66 I mi J nj R {D11Imi i =1 j =1 12 10 21 01 10 12 01 21 J nj + I mi J nj R + 2 D26 I mi J nj + I mi J nj R3 Aij . + 2 D16 I mi
00 22 4 J nj R + D22 I mi
) }
(22.4) In addition to the integrals defined in (21.108) and (21.109), this expression introduces the integrals corresponding to pq, rs = 01, 10, 12 and 21. The system of equations (21.114) or (21.115) is then modified according to the previous expression. In the case of a uniform load, on deriving the products of integrals in the form (21.116), we obtain the following system for the determination of the coefficients Aij :
2200 2002 0220 1111 2 0022 4 + D12 ( Cminj + Cminj ) + 4 D66Cminj {D11Cminj R + D22Cminj R i =1 j =1 1210 2101 1012 0121 0 0 R + 2 D26 Cminj R3 Aij = a 4 q0 I m Jn , + 2 D16 Cminj + Cminj + Cminj M N
) }
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
(22.5)
In the case of symmetric laminate plates, this system replaces the system (21.119).
22.1.2 Simply Supported Symmetric Laminate Plates

In the case of a rectangular plate simply supported along its four edges, the boundary conditions are given by Relations (21.4) and (21.5). The expressions for the moments Mx and My are deduced from the constitutive equation (14.29), and the boundary conditions on the edges are written as: edges x = 0 and x = a :
w 0 = 0, M x = D11 2w 0 x 2 2 D16 2w 0 2w 0 D12 = 0, xy y 2
(22.6)
22.1 Symmetric Laminate Plates
469
edges y = 0 and y = b :
w 0 = 0, M y = D12 2w 0 x 2 2 D26 2w 0 2w 0 D22 = 0. xy y 2
(22.7)
In fact, there exists no functions Xm(x) and Yn(y) of the approximate solution (22.3) satisfying the conditions imposed on the moments Mx and My. This difficulty is apparent, therefore, since in the Ritz method only the conditions on the displacements have to be satisfied at the boundaries. However, the convergence of the solution (22.3) may be very slow. Thus, we choose as the functions Xm(x) and Yn(y) trigonometric functions of the form:
X m ( x) = sin m
x , a y Yn ( x) = sin n . b
(22.8) (22.9)
The boundary conditions on the displacements are satisfied. Because the conditions on the moments are not satisfied there will result a less rapid convergence of the approximate solution (22.3) toward the exact solution. In the case where M = N = 1, the approximation has the form: w 0 ( x, y ) = A11 sin x y sin , a b (22.10)
where the coefficient A11 is deduced from Relation (22.5). Hence:

A11 = 16
6 D11 + 2 ( D12 + 2 D66 ) + D22 R 4
a 4 q0
(22.11)
where R is the length-to-width ratio of the plate. We notice that the first approximation does not contain the coupling terms D16 and D26. This fact results from the vanishing of the integrals in D16 and D26 for m = n = 1. The deflection is maximum at the centre of the plate. We obtain:
w 0 max = 16
6 D11 + 2 ( D12 + 2 D66 ) + D22 R 4
a 4 q0
(22.12)
As a numerical application, we consider the case of a plate with a length twice its width (R = 2), made of the laminate already considered in Subsection 21.5.4, but oriented at 30 to the plate directions. The flexural coefficients Dij, referred to the plate directions, are then determined simply by applying the orientation transformations reported in Table 11.6. Thus:
D11 = 245.23 Nm, D22 = 245.23 Nm, D12 = 92.243 Nm, D26 = 15.824, D16 = 15.824 Nm, D66 = 94.767 Nm. (22.13)
The application of Relation (22.12) leads to a maximum deflection equal to: w 0 max (22.14) = 2.591 106 N 1m 1 . 4 a q0
470
TABLE 22.1. Convergence of the approximate solution for an orthotropic plate oriented at 30, when the number of terms increases. M=N 1 3 5 7 9 11
w 0 max a 4 q0 (106 N 1m 1 )
2.591
2.455
2.471 0.01319
2.468 0.01279
2.469 0.01297
2.469 0.01288
M x a 2 q0
0.01571 0.01220
Table 22.1 gives the numerical values obtained for M = N varying from 1 to 11. The approximation using 121 terms (M = N = 11) gives a maximum deflection equal to: w 0 max (22.15) = 2.469 106 N 1m 1 . 4 a q0 The difference with the one-term approximation is 4.9%. With a nine-term approximation (M = N = 3), the difference is reduced to 0.57%.
22.1.3 Clamped Symmetric Laminate Plates

An approximate solution of the bending problem of rectangular plates, constituted of a symmetric laminate and clamped along the edges, can be obtained by introducing the functions Xm(x) and Yn(y) considered in (21.131) and (21.132). The coefficients Amn of the approximate solution (22.3) are derived by substituting the functions (21.131) and (21.132) into the system of equations (22.5). This type of solution was considered by Ashton and Waddoups [29]. Solving the system (22.5) requires preliminary the determination of the numerical values of 01 12 the integrals I mi and I mi . These values are reported in Tables A.5 and A.6 of Appendix A for the polynomial functions, and in Tables B.4 and B.5 of Appendix B for the beam functions. As an example, we consider the case of a plate made of an orthotropic layer (a unidirectional carbon fibre composite type) of mechanical characteristics:
EL = 12 ET , GLT = 0.40 ET ,
LT = 0.32.
(22.16)
The maximum deflection, determined in the case of a square plate (R = 1) using 121 terms (M = N = 11), is reported in Figure 22.1 as a function of the angle that the material directions of the layer makes with the edges of the plate. These deflections, obtained in the case where the coupling terms D16 and D26 are taken into account, are compared with the deflections obtained when the coupling terms are neglected: D16 = D26 = 0 (the case of an orthotropic plate in the plate directions). The variation of the maximum deflection is also reported in Figure 22.2 in the case of a rectangular plate of the same material, for a plate having a length twice its width (R = 2). We observe a larger influence of the orientation of the material directions than in the case of a square plate.
22.1 Symmetric Laminate Plates
471
6 ( 10 ) 5 4 3 2 1 orthotropic laminate ( D16 and D26 = 0 )

3
D=
Deflection w0max D / q0 b4
2 12 (1 LT ET EL )
E L h3
symmetric laminate ( D16 and D26 0 )
10
20
30
40
50
60
70
80
90
Orientation ( )
FIGURE 22.1. Maximum deflection of a clamped square plate constituted of a symmetric laminate as a function of the material orientation.
1.6 ( 103 )
D=
Deflection w0max D / q0 b4
1.2
2 12 (1 LT ET EL )
E L h3
0.8
0.4
10
20
30
40
50
60
70
80
90
Orientation ( )
FIGURE 22.2. Maximum deflection of a clamped rectangular plate constituted of an orthotropic laminate as a function of the material orientation.
472
22.2 RECTANGULAR CROSS-PLY PLATES 22.2.1 General Expressions

We consider a rectangular plate with dimensions a and b, constituted of an antisymmetric [0/90]p cross-ply laminate. For this type of laminate (Table 15.3), we have:
A22 = A11 , B22 = B11 , D22 = D11 , A16 = A26 = 0, B12 = B16 = B26 = B66 = 0, D16 = D26 = 0.
(22.17)
The bending equations are given by equations (16.10) to (16.12):

A11 2u0 x 2 + A66 2u0 y 2 + ( A12 + A66 ) 2v0 3w 0 B11 = 0, xy y 3 (22.18)
( A12 + A66 )
2u0 2v 2v 3w 0 + A66 20 + A11 20 + B11 = 0, xy x y y 3
4w 3u0 3v0 4w 0 4w 0 2 2 D11 40 + D D B + + 3 = q. ( ) 12 66 11 4 2 2 3 x y x y x y We study the case of hinged edges, free in the normal directions of the edges. The boundary conditions in this case are written as: edges x = 0 and x = a :
w 0 = 0, M x = 0, v0 = 0, N x = 0,
(22.19)
w 0 = 0,
M y = 0,
u0 = 0,
N y = 0.
(22.20)
The in-plane resultants and the bending moments are deduced from the constitutive equation(14.29). Thus:
N x = A11 u0 v 2w 0 + A12 0 B11 , x y x 2
u0 2w 0 2w 0 D11 D12 , M x = B11 x x 2 y 2 N y = A12 u0 v w0 + A11 0 + B11 , x y x 2 v0 2w 0 2w 0 D12 . D 11 y y 2 y 2

2
(22.21)
M y = B11
22.2 Rectangular Cross-Ply Plates
473
The boundary conditions are thus written as: edges x = 0 and x = a :

w 0 = 0, u0 2w 0 2w 0 M x = B11 D11 D12 = 0, x x 2 y 2
(22.22) (22.23)
v0 = 0,
N x = A11
u0 v 2w 0 + A12 0 B11 = 0, x y x 2
v0 2w 0 2w 0 D12 D = 0, 11 y y 2 y 2
w 0 = 0, M y = B11
(22.24) (22.25)
u0 = 0,
N y = A12
u0 v 2w 0 + A11 0 + B11 = 0. x y x 2
The transverse load q = q(x, y) is expanded in a double Fourier series (Relation (21.8)): q ( x, y ) =
qmn sin m a sin n b .

m =1 n =1
(22.26)
Next, the solutions of the bending problem are obtained by expressing the displaycements in the form of double Fourier series satisfying the boundary conditions (22.22) to (22.25). That is: u0 = v0 = w0 =
Amn cos m a sin n b , Bmn sin m a cos n b ,

m =1 n =1 m =1 n =1
y y
(22.27)

m =1 n =1
Substituting Expressions (22.27) into the system (22.18), we obtain:
( m2 A11 + n2 R 2 A66 ) Amn + mnR ( A12 + A66 ) Bmn m3 B11Cmn = 0,

a mnR ( A12 + A66 ) Amn + ( m 2 A66 + n 2 R 2 A11 ) Bmn + n3 R3 m B11 Amn + n R B11Bmn a4 4 4 4) 2 2 2 ( m n R D D D m n R C qmn . + + + 2 + 2 = ( ) mn 11 12 66 a 4 The determinant of this system of equations is:
3 3 3
B11Cmn = 0,
(22.28)
mn ,
(22.29)
474
with
4 4 4 2 2 2 mn = ( m + n R ) D11 + 2m n R ( D12 + 2 D66 ) 2 2 2 2 2 2 2 2 2 2 ( m A11 + n R A66 )( m A66 + n R A11 ) m n R ( A12 + A66 ) 2 2 2 2( 4 4 4 4 4 4 8 8 8 B11 m n R m + n R ) A11 + 2m n R ( A12 + A66 ) + ( m + n R ) A66 .
(22.30) Solving the system (22.28) leads to:

Amn = ma 3 B11
mn
3
4 2 2 2 4 4 qmn m A66 + m n R A11 + n R ( A12 + A66 ) , 4 2 2 2 4 4 qmn m ( A12 + A66 ) + m n R A11 + n R A66 ,
(22.31) (22.32)
Bmn = Cmn =
na 3 RB11
mn
a4
4
mn
4 2 2 2 2 2 4 4 qmn m A11 A66 + m n R ( A11 A12 2 A12 A66 ) + n R A11 A66 .
(22.33) The expressions for the resultants, moments and stresses can next be derived from Expressions (16.13) to (16.25). In the case of a laminate such that B11 = 0, the solution reduces to:
u0 = 0,
w0 = v0 = 0,
a4
qmn
m =1 n =1
x y sin n (22.34) a b . ( m4 + n 4 R 4 ) D11 + 2m2 n2 R 2 ( D12 + 2 D66 ) sin m
This expression is the same as that for the deflection (21.14) obtained in the case of a simply supported plate made of an orthotropic laminate such that D22 = D11.
22.2.2 Influence of the Moduli

In the case of a loading of the form: q = q0 sin x y sin , a b (22.35)
the solutions (22.27) of the bending problem reduce to: x y sin , a b x y v0 = B sin cos , a b x y w 0 = C sin sin . a b u0 = A cos
(22.36)
475
The amplitudes A, B and C are expressed as:

A= a3 B11
D
3
4 2 4 q0 A66 (1 + R ) + A11R + A12 R , 2 4 q0 A12 + A11R + A66 (1 + R ) ,
B= C= with
a3 B11R
D
4
(22.37)
a4
4 2 2 2 q0 A11 A66 (1 + R ) + ( A11 A12 2 A12 A66 ) R ,
4 2 2 2 D= D11 (1 + R ) + 2 ( D12 + 2 D66 ) R ( A11 + A66 R )( A66 + A11R ) 2 2 2 4 4 8 B11 ( A12 + A66 ) R 2 . A11R (1 + R ) + 2 ( A12 + A66 ) R + A66 (1 + R )
It results that in the case of a square plate (R = 1), the maximum deflection can be written in the form:
w 0 max =
a 4 q0
1 2 ( D11 + D12 + 2 D66 )
1
2 B11 1 D11 + D12 + 2 D66
(22.38)
with
A11 + A12 + 2 A66 2 A11 + 2 ( A11 A12 ) A66
2 A12
(22.39)
where the term is independent of the number of layers. For an antisymmetric [0/90]p cross-ply laminate, the stiffness coefficients are deduced from the results reported in Table 15.3. Thus:
1 E A11 = 1 + T 2 EL A66 = Q66 h, Q11h, A12 = Q12 h, 1 ET B11 = 1 Q11h 2 , 8 p EL
(22.40) (22.41)
(22.42)
1 E D11 = 1 + T 2 EL with Q11 = EL

2 1 LT
3 Q11h 12 ,
D12 =
Q12 h3 , 12 ,
D66 =
Q66 h3 , 12
ET EL
Q12 =
LT ET
2 1 LT
ET EL
Q66 = GLT .
(22.43)
The maximum deflection can thus be put in the form:

w 0 max = 12 a q0
4 3 4 2 1 LT
ET EL
ET h 1 + EL + 2 + 4 1 2 ET GLT LT LT ET EL ET
E 1 f L , 2 , (22.44) ET p
476
with
1 E f L , 2= ET p 1 3 16 p 2 1 EL 2 1 GLT E + 2 + E 4 T T . 2 EL 1 E T GLT EL 1 2 E + 4 + (1 2 LT ) E LT T T (22.45) 1
Figure 22.3 shows the variation of the maximum deflection as a function of the modulus ratio EL/ET, in the case where the layers have the following mechanical characteristics: GLT = 0.50, LT = 0.30. (22.46) ET The results reported show that the solution for the orthotropic laminate (B11 = 0) is reached rapidly when the number of layers of the laminate is increased.
0.025
0.020
0.015 [0/90] 0.010 [0/90]3 [0/90]2 0.005 orthotropic laminate (B11 = 0) 0 0 10 20 30 40 50
Modulus ratio EL /ET
FIGURE 22.3. Maximum deflection as a function of the modulus ratio for a square plate made of an antisymmetric cross-ply laminate.
477
22.2.3 Influence of the Length-to-Width Ratio

In the case of a loading of the form (22.35), the maximum deflection is, from (22.37), given by:
w 0 max =
a 4 q0 2 2 A A (1 + R 4 ) + ( A11 A12 2 A12 A66 ) R 2 . 4 11 66 D
(22.47)
In the case of an orthotropic laminate (B11 = 0), the maximum deflection is written as:
w 0 max =
a 4 q0
D11 (1 + R 4 ) + 2 ( D12 + 2 D66 ) R 4

a 4 q0 b 4 q0
(22.48)
and in the case of a high length-to-width ratio of the plate, it reduces to:
w 0 max =
4 D11R 4
4 D11
(22.49)
This expression is the same as Relation (19.37) obtained in the case of a cylindrical bending (the roles of a and b being interchanged in the two cases). For an antisymmetric [0/90]p cross-ply laminate, the stiffness coefficients of which are expressed in (22.40) to (22.43), Relation (22.49) can be put in the form:
w 0 max = b 4 q0
R4 1 , ET h3 D( R) 1 E ( R) p2 1
(22.50)
with
D( R) = d11 (1 + R 4 ) + 2 ( d12 + 2d66 ) R 2 , E ( R) =

2 b11 A( R) , 64 B( R)
(22.51)
2 2 B ( R) = a11a66 (1 + R 4 ) + ( a11 a12 2a12 a66 ) R 2 ,
A( R) = a11 (1 + R 4 ) R 2 + 2 ( a12 + a66 ) R 4 + a66 (1 + R8 ) ,
and ET 1 1 LT EL EL , d12 = , d11 = 24 1 2 ET ET 12 1 2 ET LT LT EL EL a11 = 12d11, a12 = 12d12 , a66 = 12d66 , ET 1 EL 1 EL b11 = . 8 1 2 ET ET LT EL 1+ 1 GLT , 12 ET (22.52)
d66 =
478
0.04
Deflection w0max ET h3 / q0 b4
0.03
[0/90]
[0/90]2 0.02
[0/90]3
0.01
orthotropic laminate (B11 = 0)
Length-to width ratio a /b

FIGURE 22.4. Maximum deflection as a function of the length-to-width ratio of a rectangular cross-ply laminate plate.
In the case of an orthotropic laminate and cylindrical bending, Relation (22.51) leads to: 1 b 4 q0 . (22.53) w 0 max = 4 d11 ET h3 The influence of the length-to-width ratio of the plate is reported in Figure 22.4, in the case where the layers have the characteristics (22.47) and for the modulus ratio EL/ET = 15. As previously, we observe that the solution for the orthotropic laminate is rapidly reached when the number of layers of the antisymmetric cross-ply laminate is increased. Figure 22.4 also shows that the solution for the cylindrical bending (Chapter 19) is obtained for length-to-width ratios practically equal to 4.
22.3 RECTANGULAR ANGLE-PLY PLATES

In this section, we consider the case of a rectangular plate made of a []p angle-ply laminate, having an even number nc = 2p of layers with a total thickness h. The laminate is alternated and antisymmetric, and the stiffness coefficients
22.3 Rectangular Angle-Ply Laminates
479
are (Relations (15.26)) expressed as: Aij = hQij Aij = 0 Bij = 0 Bij = h2 Qij 4p si ij = 11, 12, 22, 66, si ij = 16, 26. si ij = 11, 12, 22, 66, si ij = 16, 26. si ij = 11, 12, 22, 66, si ij = 16, 26 and n even. (22.54)
h3 Dij = Qij 12 Dij = 0
The bending equations are deduced from Equations (16.1) to (16.3) and may be written as:
A11 2 u0 x 2 + A66 2u0 y 2 + ( A12 + A66 ) 2v0 3w 3w 0 3B16 2 0 B26 = 0, xy x y y 3
( A12 + A66 )
D11 4w 0 x 4
2u0 2v 2v 3w 0 3w 0 3 B + A66 20 + A22 20 B16 = 0, 26 xy x y x3 xy 2 4w 0 x 2y 2 + D22 4w 0 y 4
(22.55)
+ 2 ( D12 + 2 D66 )
3u0 3u0 3v0 3v0 3 B + 3 B16 + = q. x 2y x3 26 y 3 xy 2
We study the case of hinged edges, free in the tangential directions of the edges. The boundary conditions are then: edges x = 0 and x = a :
w 0 = 0,
M x = 0,
u0 = 0,
N xy = 0,
(22.56)
w 0 = 0,
M y = 0,
v0 = 0,
N xy = 0.
(22.57)
The in-plane resultant Nxy and the bending moments are deduced from the constitutive equation (14.29). That is:
v 2w 0 2w 0 u N xy = A66 0 + 0 B16 B , 26 x x 2 y 2 y v 2w 0 2w 0 u , M x = B16 0 + 0 D11 D 12 x x 2 y 2 y 2w 0 2w 0 u0 v0 . M y = B26 + D12 D22 x x 2 y 2 y
(22.58)
480
The boundary conditions are then written as: edges x = 0 and x = a :

w 0 = 0,
u0 = 0, v 2w 0 2w 0 u M x = B16 0 + 0 D11 D 12 = 0, x x 2 y 2 y v 2w 0 2w 0 u N xy = A66 0 + 0 B16 B = 0, 26 x x 2 y 2 y 2w 0 2w 0 u0 v0 M y = B26 + D12 D22 = 0, x x 2 y 2 y
(22.59) (22.60)
w 0 = 0, v0 = 0,
(22.61)
v 2w 0 2w 0 u N xy = A66 0 + 0 B16 B = 0. (22.62) 26 x x 2 y 2 y The transverse load q = q(x, y) is expanded in a double Fourier series, and the solution of the bending problem is obtained by expressing the displacements in the form of a double Fourier series satisfying the boundary conditions (22.59) to (22.62). So:
u0 = v0 =
Amn sin m a cos n b , Bmn cos m a sin n b ,

m =1 n =1 m =1 n =1
y y
(22.63)
w0 =

m=1 n =1
By substituting Expressions (22.63) into the bending equations (22.55), we obtain:
( m2 A11 + n2 R 2 A66 ) Amn + mnR ( A12 + A66 ) Bmn nR ( 3m 2 B16 + n 2 R 2 B26 ) Cmn = 0,
mnR ( A12 + A66 ) Amn + ( m A66 + n R A22 ) Bmn
2 2 2
m ( m 2 B16 + 3n 2 R 2 B26 )
Cmn = 0,
(22.64)
nR ( 3m 2 B16 + n 2 R 2 B26 ) Amn m ( m 2 B16 + 3n 2 R 2 B26 ) Bmn
a3 4 2 2 2 4 4 2 2 + m D + D + D m n R + n R D C = qmn . ( ) 11 12 66 22 mn a 3
The determinant of this system of equations is:
mn ,
(22.65)
22.3 Rectangular Angle-Ply Laminates
481
with
4 2 2 2 4 4 mn = m D11 + 2m n R ( D12 + 2 D66 ) + n R D22 2 2 2 2 2 2 2 2 2 2 ( m A11 + n R A66 )( m A66 + n R A22 ) m n R ( A12 + A66 )
+ 2m 2 n 2 R 2 ( A12 + A66 ) + ( 3m 2 B16 + n 2 R 2 B26 )( m 2 B16 + 3n 2 R 2 B26 ) m 2 ( m 2 B16 + 3n 2 R 2 B26 )
( m2 A11 + n2 R 2 A66 ) 2 n 2 R 2 ( 3m 2 B16 + n 2 R 2 B26 ) ( m 2 A66 + n 2 R 2 A22 ) .

2
(22.66)
Solving the system (22.64) then leads to:
Amn =
nRa3
mn
ma3
2 2 2 2 2 2 qmn ( m A66 + n R A22 )( 3m B16 + n R B26 )
m 2 ( A12 + A66 ) ( m 2 B16 + 3n 2 R 2 B26 ) ,

2 2 2 2 2 2 qmn ( m A11 + n R A66 )( m B16 + 3n R B26 )
Bmn =
mn
a4
n 2 R 2 ( 3m 2 B16 + n 2 R 2 B26 ) ( A12 + A66 ) ,

2 2 2 2 2 2 qmn ( m A11 + n R A66 )( m A66 + n R A22 ) 2 m 2 n 2 R 2 ( A12 + A66 ) .
(22.67)
Cmn =
mn
The expression of coefficient Cmn can be written in the form:

0 Cmn = Cmn
1 + ijklpq
1 , Aij Bkl B pq
0 mn
(22.68)
with
0 Cmn =
a4
4
0 mn
qmn , (22.69)
0 mn = m 4 D11 + 2 ( D12 + 2 D66 ) m 2 n 2 R 2 + n 4 R 4 D22 ,
i, j , k , l , p, q = 1, 2, 6. When the number of layers is large enough, Relations (22.54) show that:
0 Cmn Cmn ,
(22.70)
and the expression for the deflection is the same as the solution for an orthotropic laminate (B16 = B26 = 0), so by (22.54): x y sin m sin n 4 12 a a b qmn 4 w0 = 4 3 . (22.71) + 2m 2 n 2 R 2 ( Q12 + 2Q66 ) + n 4 R 4Q22 m Q11 h m=1 n=1
482
0.010
0.008
[ ] [ ]2
0.006
[ ]3
0.004
orthotropic laminate (B16 and B26 = 0)
0.002
10
15
20
25
30
35
40
45
Orientation ( )
FIGURE 22.5. Maximum deflection of a square angle-ply plate as a function of the angleply orientation.
As in the case of cross-ply laminates (Section 22.2), the orthotropic solution is the limit solution when the number of layers increases. It results that the coupling owed to the terms B16 and B26 can be important for a small number of layers, it decreases rapidly when the number of laminate layers increases. These results are illustrated in Figure 22.5, in the case of a square plate subjected to a transverse load of the form: x y (22.72) q = q0 sin sin . a b
The maximum deflection at the plate centre is reported as a function of the orientation of the layers, in the case where the layers have the following mechanical characteristics: EL = 15, ET GLT = 0.50, ET
LT = 0.30.
(22.73)
EXERCISES
22.1 An orthotropic plate of length twice its width is clamped along the four edges. The material directions make an angle of 30 with respect to the plate edges. The flexural stiffnesses referred to the material directions are:
0 0 D12 = 0.09 D11 , 0 0 D66 = 0.125 D11 , 0 0 D22 = 0.75 D11 .
Exercises
483
Derive the flexural stiffnesses in the plate directions. Using the beam functions, establish the system (22.5) for M = N = 3.
0 ). Calculate the coefficient Aij (referred to a 4 q0 D11 Express the deflection at point (x, y), then at the centre of the plate.
22.2 Do Exercise 22.1 again in the case where the plate has two adjacent edges clamped, the other two being free. The problem will be solved by using the beam functions considered in Exercise 21.5.
CHAPTER 23
Buckling of Laminate or Sandwich Beams and Plates
23.1 FUNDAMENTAL RELATIONS INCLUDING BUCKLING 23.1.1 Introduction

In the case f symmetric or orthotropic laminates, the equations of the classical laminate theory (16.1) to (16.3) or the theory taking transverse shear into account (17.27) to (17.31) are uncoupled: the equations that determine the transverse displacement of the laminate are independent of equations for the in-plane displacements. It results from these equations that an in-plane loading (a displacement or an applied force) can produce only in-plane deformations. For example, if we consider a straight beam (Figure 23.1), clamped at one of its ends and subjected to a compressive force F, the plate equations show that the beam shorten, the mean line staying straight (Figure 23.1a). If we create a small transverse perturbation at a point M of the beam, the system recovers its initial position when the perturbation disappears: the equilibrium is stable. z M z (a) Fcr (b)
FIGURE 23.1. Buckling of a beam.
23.1 Fundamental Relations Including Buckling
485
If the load F is increased progressively, it is nevertheless observed that, in the case of a beam with high aspect ratio and for a certain critical value Fcr corresponding to a value cr of the normal stress clearly below the fracture stress, it is induced a transverse displacement (Figure 23.1b). It is then no longer possible to increase the compressive load without initiating fracture or a large deflection of the beam. This phenomenon is referred as buckling or elastic instability. In order to describe this phenomenon, it is then necessary to develop equations that take into account a transverse displacement of plates subjected to in-plane loading.
23.1.2 Plate Equations Taking Account of Buckling

To take into account the phenomenon of buckling, it is thus necessary to write the plate equations that consider transverse deformation, by introducing for each point of the structure the coordinates after deformation in contrast to the initial theory. We establish these equations by considering the in-plane resultants of the actions applied to a plate element of sides d x and d y (Figure 23.2). The transverse deformation in the plane (x, z) under the action of the resultant Nx is represented schematically in Figure 23.3. It results that the z component of the load Nx, exerted on the plate element is, for small deformations:
2 1 N x w 0 w 0 w 0 N d x d y d x N d y + + x . x x d x d y x x x 2
(23.1)
z
Nxy d y
Nx d y
Nxy d x Ny d x
dx
dy
Ny y
( Ny +
d y) d x
( Nxy +
Nxy y
d y) d x
Nx ( Nx + d x) d y y
( Nxy +
Nxy y
d x) d y
FIGURE 23.2. In-plane resultants applied to a plate element.
486
Chapter 23 Buckling of Laminate or Sandwich Beams and Plates
y ( Nx +
dx
Nx d x) d y x
w0 w0 ) dx + ( x x x x
Nx d y
w0 x
FIGURE 23.3. In-plane normal resultants acting on an element of a deformed plate.
Restricting to the two-order terms, the resultant per unit area of the plate in the z direction is:
Nx
2w 0 x
2
N x w 0 . x x
(23.2)
Similarly, the z component due to the resultant Ny is:

Ny 2w 0 y 2 + N y w 0 . y y
(23.3)
The z component due to the shear resultant Nxy can be evaluated from the diagram of Figure 23.4. It is expressed as:
N xy N xy w 0 2w 0 w 0 2w 0 1 d d dy N + x + y + N + + xy dx xy y d x d y x xy y xy x w 0 w 0 d y N xy d x . (23.4) + N xy y x Therefore on restricting to the two-order terms, this component is written as:
2 N xy 2w 0 N xy w 0 N xy w 0 . + + xy x y y x
(23.5)
By regrouping Expressions (23.2), (23.3) and (23.5), the total z component is: Nx 2w 0 x 2 + 2 N xy 2w 0 2w 0 + Ny xy y 2 w 0 N y N xy + + x y y .
N xy w N + 0 x + x x y
(23.6)
The plate equations (13.56) and (13.58) show that the last two terms of the preceding expression are zero in the case of problems of statics and are of
487
dy
Nx y d y Nx y d x
dx
y w0 y
( Nx y +
Nx y y
w0 x
d y) d x
w0 w0 ) dy + ( x y x ( Nx y +
w0 w0 ) dx + ( y x y
Nx y x
d x) d y
FIGURE 23.4. In-plane shear resultant acting on an element of a deformed plate.
third-order in the case of problems of dynamics. It results that the z component (23.6) reduces to:
Nx
2w 0 x 2 + 2 N xy 2w 0 2w 0 . + Ny xy y 2
(23.7)
The plate equations taking account of the transverse deformation are then obtained by introducing the z component into Equations (13.56). Whence: N x N xy 2u + = s 20 , x y t N y y
Nx
2w 0 x 2 + 2 N xy
(23.8) (23.9) (23.10)
N xy x
= s
2v0 t 2
2w 0 2w 0 Qx Q y 2w 0 , q + Ny + + + = s xy x y y 2 t 2 M x M xy + Qx = 0, x y M y y + M xy x Qy = 0.
(23.11) (23.12)
These equations differ from the classical equations (13.56) solely by the modification of the third equation. These equations can also be written by eliminating
488
the shear resultants in a form analogous to (13.57). Thus:
N x N xy 2u0 + = s 2 , x y t N y y
2M x x 2 + 2M y y 2 +2 2 M xy xy
2
(23.13) (23.14)
N xy x
= s
2v0 t 2
+ Nx
2w 0 x 2
2 2
+ 2 N xy
w0 w0 w0 + Ny + q = s . 2 xy y t 2
(23.15)
These equations differ from Equations (13.57) by the modification of the third equation.
23.1.3 Equations of the Classical Laminate Theory Taking Account of Transverse Displacement
If Equation (23.15) is used in this form to determine the critical buckling load, the problem then becomes nonlinear in the case of nonsymmetric laminates (Bij 0). In order to solve the buckling problem, it is so necessary to use a perturbation method that allows us to derive linear equations. Thus, the displacement field is written in the form:
i u0 = u0 + u0 , i v0 = v0 + v 0 , i w0 = w0 + w 0 ,
(23.16)
i i i , v0 , w0 ) is the prebuckling displacement field, (u0 , v0 , w 0 ) is an where (u0 admissible and arbitrary displacement field (satisfying the boundary conditions and the continuity), is an infinitesimal quantity independent of the coordinates. The buckling phenomenon thus appears as the process which produces an infinitesimal change in the equilibrium position. Relation (23.16) associated to the constitutive equation (14.30) leads to the matrix relations:
N = A im + B i + ( A m + B ) = Ni + N , M = B im + D i + ( B m + D ) = M i + M .
(23.17)
Substituting Relations (23.16) and (23.17) into Equation (23.15), we obtain a firstdegree equation in , by neglecting the second-degree terms in . This equation is satisfied whatever the value of , if the terms of the equation in are zero. Thus, we obtain the equations:
489
i 2M x
x 2
2M x x 2
2M i y y 2
My y 2
2
+2
i 2 M xy
xy
2
i Nx
i 2w 0
x 2
2w 0 x 2
i 2 N xy
i 2 i i 2w 0 2w 0 i w0 i + Ny + q = s , xy y 2 t 2
(23.18)
+ +2 M xy xy
i + Nx
+ Nx
i 2w 0
x 2
i + 2 N xy
i i 2w 0 2w 0 2w 0 2w 0 2w 0 . N q + 2 N xy + Ni + + = y y s xy xy y 2 y 2 t 2
(23.19) The first equation (23.18), which is the same as (23.15), allows us to determine the initial configuration (the elastic configuration) in the case of large transverse deformations. This equation is, however nonlinear, and the usual simplification consists to determine the initial configuration considering the linear theory (Equations (13.57)). Because the initial configuration is determined in the case of small i transverse displacements w 0 , it follows that the terms containing the initial curvatures can be neglected in Equation (23.19). This equation then becomes:
2M x x 2 + 2M y y 2 +2 2 M xy xy
i + Nx
2w 0 x 2
i + 2 N xy
2w 0 2w 0 2w 0 + Ni + q = . y s xy y 2 t 2 (23.20)
This equation constitutes the equation of buckling which is associated to Equations (23.13) and (23.14), unchanged with respect to Equations (13.57). The fundamental equations of the classical laminate theory, allowing us to formulate buckling, thus comprise Equations (16.1) and (16.2). Next, taking (23.20) into account, Equation (16.3) is replaced by:
D11
4w 0 x 4 3u0 x3
+ 4 D16 3B16
4w 0 x3y 3u0 x 2y
+ 2 ( D12 + 2 D66 ) ( B12 + 2 B66 ) 3v0 xy 2
4w 0 x 2y 2
+ 4 D26 3u0 y 3
4w 0 xy 3 B16
+ D22 3v0 x3
4w 0 y 4
B11
3u0 xy 2 y 3
B26 + s
( B12 + 2 B66 )
i = Nx
3v0 x 2y
3B26
B22
3v0
2w 0 t 2
(23.21)
2w 0 x 2
i + 2 N xy
2w 0 2w 0 + Ni +q. y xy y 2
In the case of symmetric laminates (Bij = 0), the in-plane equations and the flexural equations are uncoupled, with in the case of pure bending: u0 = v0 = N x = N y = N xy = 0 . (23.22)
490
It is then no longer necessary to distinguish the prebuckling and buckling equations. In this case, Equation (23.21) is written:
D11
4w 0 x 4
+ 4 D16
4
4w 0 x3y
4
+ 2 ( D12 + 2 D66 ) w0 t
2 2
4w 0 x 2y 2
+ 4 D26
2
4w 0 xy 3
2
+ D22
w0 y
+ s
= Nx
w0 x
2
+ 2 N xy
w0 w0 + Ny + q. xy y 2
(23.23)
In the case of orthotropic laminates (D16 = D26 = 0), Equation (23.23) reduces to:
D11 4w 0 x 4 + 2 ( D12 + 2 D66 ) 4w 0 x 2y 2 x 2 + D22 4w 0 y 4 + s 2w 0 t 2
= Nx
2w 0
+ 2 N xy
2w 0 2w 0 + Ny + q. xy y 2
(23.24)
Lastly, in the case of an isotropic homogeneous plate, we have (Relations (15.2)): D11 = D22 = 2 ( D12 + 2 D66 ) = D , (23.25) with
D= h3 , 1 2 12 E
(23.26)
and Equation (23.24) of buckling can be written according to the classical equation of isotropic homogeneous plates:
D w0 +
4
2w 0 t 2
= Nx
2w 0 x 2
+ 2 N xy
2w 0 2w 0 + Ny +q, xy y 2
(23.27)
introducing the operator

=
4
4 x 4
4 x 2y 2
4 y 4
(23.28)
23.1.4
Energy Formulation of the Buckling Problem
In the case where the transverse deformation is taken into account, the actions exerted on the laminate result from the transverse loads q applied to the lower and upper faces of the laminate, and from the in-plane loads. The variation of the work of the actions exerted on the laminate is thus written as:
W = Wf + Wm ,
(23.29)
491
where the variation Wf has been expressed in (16.44). The energy function Wm of the in-plane actions induced by a deflection w0 is expressed as:
Wm =
i i + Ni )dx dy , Nx xx ( y yy + N xy xy
(23.30)
, are the in-plane strains resulting from the deflection w0. where xx yy and xy
These strains are deduced from the general expressions (6.10), including the large deformations. When only large transverse deformations are considered, the strains can be written as:
xx = yy xy
u0 1 w 0 1 w 0 0 + = xx + , x 2 x 2 x
2 2
v 1 w 1 w 0 = 0 + 0 =0 , yy + y 2 y 2 y u v w w 0 w w 0 0 = 0+ 0+ 0 = xy + 0 . y x x y x y
(23.31)
It results that the in-plane strains owed to the transverse deformation are expressed as:
1 w 0 = xx , 2 x
2
1 w 0 , yy = 2 y
= xy
w 0 w 0 , x y
(23.32)
and the energy Wm is: Wm = 1 2
2 i w 0 2 i w 0 i w 0 w 0 N N + + 2 N xy x d x d y . (23.33) y x y x y
The variation of the work done by the in-plane loads is finally given by taking the variation of (23.33), that is: Wm =
2 2 i 2w 0 i w0 i w0 + + N N 2 N x w0 d x d y . y xy xy x 2 y 2
(23.34)
The variational formulations of the buckling problem are deduced from Relations (8.45) in the case of statics problems and from (8.57) in the case of dynamics problems. Here they may be written: in the case of problems of statics: (U d Wf Wm ) = 0 , in the case of problems of dynamics: (23.35)
t1 t0
(U d Wf Wm Ec ) d t = 0 ,
(23.36)
where the functions Ud, Wf, Wm and Ec are given respectively by Expressions (16.35) to (16.38), (16.45), (23.33) and (16.42).
492
23.1.5 Equations of Theory with Transverse Shear Taking Account of Transverse Displacement
A development similar to that of Subsection 23.1.3, applied to Equation (23.10), leads to the buckling equation:
2 2 2 Qx Qy 2w 0 i w0 i w0 i w0 + + Nx + N + N + q = 2 . xy y s x y xy x 2 y 2 t 2
(23.37)
The fundamental equations of the laminate theory with transverse shear, allowing us to formulate buckling, are thus constituted of Equations (17.27) to (17.31), Equation (17.29) being replaced, on account of (23.37), by the equation:
x 2w 0 x y y 2w 0 2w 0 H 55 H 2 H + + + + + + 45 44 x x xy x 2 y 2 y y + q+
i Nx
2w 0 x
i + 2 N xy 2
2 2w 0 2w 0 i w0 , + Ny = s xy y 2 t 2
(23.38)
where the coefficients Hij are the transverse shear coefficients introduced in Relation (17.53).
23.1.6 Equations of Sandwich Theory Taking Account of Transverse Displacement

The buckling equation of sandwich plates is identical to Equation (23.37). Considering the elements reported in Subsection 18.3.2, the fundamental equations of sandwich plates taking account of buckling consist of Equations (17.27), (17.28), (17.29), (18.31) and (18.32), with Relation (17.29) being modified as follows:
y 2w 0 2w 0 2w 0 c x c y c x + hC55 hC 2 hC + + + + + 45 44 x x xy x 2 y 2 y y + q+
i Nx
2w 0 x
i + 2 N xy 2
2 2w 0 2w 0 i w0 . + Ny = s xy y 2 t 2
(23.39)
When the material directions of the core coincide with the plate directions, the c are given by: coefficients Cij
c c = G23 C44 , c c = G13 C55 ,
c = 0, C44
(23.40)
and Relation (23.39) reduces to:
23.2 Buckling Under Cylindrical Bending
493
2w 0 2w 0 c x c y + + + hG13 hG 23 x y x 2 y 2
i + q + Nx
2w 0 x
2
i + 2 N xy
2w 0 2w 0 2w 0 + Ni = . y s xy y 2 t 2
(23.41)
23.2 BUCKLING UNDER CYLINDRICAL BENDING 23.2.1 Classical Laminate Theory

We consider the case of a laminate plate, made of an arbitrary number of layers and that is very long in the y direction (Section 19.1). This plate is subjected to an i initial in-plane compressive load: N x = N 0 (Figure 23.5), and no transverse load is applied: q = 0. Equation (19.4) is replaced by the equation:
D11 4w 0 x 4 B11 3u0 x3 B16 3v0 x3 + N0 2w 0 x 2 + s 2w 0 t 2 = 0,
(23.42)
and reduces for static buckling to:

D11 d 4w 0 d x4 B11 d3u0 d x3 B16 d3v0 d x3 + N0 d 2w 0 d x2 =0.
(23.43)
On differentiating Equations (19.5) and (19.6), and substituting the results into Equation (23.43), we obtain the differential equation w0:
d 4w 0 d x4 + A d 2w 0 N0 =0, D d x2
(23.44)
where the coefficients A and D are expressed by (19.7) and (19.9). The buckling equations are then constituted of Equations (19.5), (19.6) and (23.44). In the case where the plate is simply supported along the edges x = 0 and x = a, the boundary conditions are satisfied taking the displacements as:
x , a x v0 = Bm cos m , a x w 0 = Cm sin m . a u0 = Am cos m
(23.45)
494
i Nx = N0
FIGURE 23.5. A plate of great length subjected to compression.
On substituting these expressions into Equations (19.5), (19.6) and (23.44), we obtain the equations: m A Am Cm = 0, a B m C Bm Cm = 0, (23.46) a A m 2 2 A 2 N 0 Cm = 0. D a A nonzero solution is obtained in the case where:
m 2 2 a
2
A N0 = 0 . D
(23.47)
Thus the critical load corresponding to mode m is:

N cr = m 2
2 D
a2 A
(23.48)
It results that the effective critical load is the smallest load, corresponding to the fundamental mode fundamental (m = 1) with a half wave deformed shape:
N cr =
2 D
a2 A
(23.49)
Equations (23.46) show that the deflection w0 then has an arbitrary amplitude Cm. It results that the equilibrium obtained at critical loading is unstable. In the case where there exists no in-plane flexural coupling (Bij = 0), the critical load is, by(19.9), given by:
495
= N cr
2
a2
D11 .
(23.50)
In the presence of coupling, the critical load is thus written as: , N cr = (1 H ) N cr with B B + B16C H = 11 . AD11
(23.51) (23.52)
As the coefficient H is positive, it results that the in-plane flexural coupling reduces the critical buckling load.
23.2.2 Effect of Transverse Shear

We consider again the compression loaded plate studied in the preceding subsection in the case of an orthotropic laminate:
A16 = A26 = 0, B16 = B26 = 0, D16 = D26 = 0.
(23.53)
With the assumptions made in (19.39) and taking into account Equation (23.38), Equations (19.40) to (19.42) are modified as follows:
A11
2u0 x 2 + B11 2 x x 2 = s 2u0 t 2 +R 2 x t 2
(23.54)
2 2w 0 2w 0 i w0 k55 F55 x + q N + + = , x s x x 2 x 2 t 2
(23.55)
B11
2 u0 x 2
+ D11
2 x x 2
w 2u 2 x k55 F55 x + 0 = R 20 + I xy . (23.56) x t t 2
i In the case of buckling under compression ( N x = N 0 ) and of a symmetric laminate (Bij = 0), the preceding equations are simplified as:
u0 = 0 ,
(23.57)
d d 2w 0 d 2w 0 k55 F55 x + N = 0, 0 dx d x2 d x2
D11 d 2 x
2
(23.58) (23.59)
d w0 k55 F55 x + = 0. dx dx
In the case of simple supports, the boundary conditions reduce to: w 0 = 0, M x = D11 d x = 0. dx (23.60)
A solution for Equations (23.58) and (23.59), satisfying the boundary conditions
496
(23.60) has the form:
x = Am cos m
x , a x w 0 = Bm sin m . a
(23.61)
Substituting these expressions into Equations (23.58) and (23.59), we obtain:
a2 a k55 F55 Bm = 0, D11 + 2 2 k55 F55 Am + m m a k55 F55 Am + ( k55 F55 N 0 ) Bm = 0. m
(23.62)
A nonzero solution exists in the case where the determinant of the matrix of the coefficients Am, Bm is zero, which leads to:
N0 = m 2 2 k55 F55 D11 m 2 2 D11 + a 2 k55 F55
(23.63)
The critical load, which corresponds to the smallest value of N0, is obtained for the fundamental mode m = 1:
N cr =
2 k55 F55 D11 2 D11 + a 2 k55 F55

1 1 + 2S
(23.64)
The critical buckling load can be put into the form:

N cr =
, N cr
(23.65)
where S is the term taking account of the transverse shear expressed by Relation is the critical buckling load neglecting the transverse shear and (19.52), and N cr given by Expression (23.50). Expression (23.65) shows that the shear deformation decreases the critical buckling load. The effect of the transverse shear (19.55) depends on the modulus ratio Q11 G13 and the length-to-thickness ratio a/h. In the case of a symmetric laminate [0/90/90/0] made of unidirectional layers containing 60 % of glass fibres and with characteristics:
EL = 46 GPa, GLT = 4.6 GPa, ET = 10.5 GPa, GTT = 4 GPa,
LT = 0.3,
(23.66)
the modulus ratio Q11 G13 is close to 10. The variation of the critical buckling load as a function of the length-to-thickness ratio a/h is reported in Figure 23.6 for k55 = 1 and k55 = 2 . For values of the length-to-thickness greater than 20, the 3 solution (23.65) is practically similar as the value deduced from the classical laminate theory.
497
1.2
Critical buckling load N cr N cr
classical theory
1.0
0.8
k55 = 1 k55 = 2/3
with transverseshear
0.6
Q11 = 10 G13 = N cr
2
a2
D11
40
0.4
10
20
30
FIGURE 23.6. Critical buckling load of a long plate loaded under compression as a function of the length-to-thickness ratio.
Length-to-width ratio h
23.2.3 Buckling of a Sandwich Plate

In the case of a plate made of sandwich material, under cylindrical bending as considered in Section 19.6, and subjected to an in-plane compression N0 (Figure 23.5), Equations (19.101) and (19.102) are modified as follows:
D11 d 2 x
2
d w0 hG13 x + =0, dx dx
(23.67) (23.68)
d x d 2w 0 d 2w 0 hG13 + =0. N0 dx d x2 d x2
These equations have the same form as Equations (23.58) and (23.59). By transposition of the results obtained, the critical buckling load is expressed in the form (23.65), with
N cr =
2
a2
D11 ,
(23.69)
and S= 1 D11 . ha 2 G13 (23.70)
The variation of the critical load as a function of the length-to-thickness ratio a/h has the same form (Figure 23.6) as in the case of a laminate plate when the transverse shear effect is considered.
498
23.3 BUCKLING OF BEAMS 23.3.1 Buckling Equation

In the framework of beam theory (Section 20.2), Equation (23.20) is written in the form: 2 2M x 2w 0 i w0 + = N . (23.71) x s x 2 x 2 t 2 The moment and the deflection are linked by Expression (20.10), thus: d 2w 0 dx
2
M Mx = b , Ex I Ex I
(23.72)
where Ex and I are respectively the bending modulus and moment of the beam introduced in (20.11) and (20.12). Equation (23.71) may be written as:
2 Ex I 4w 0 2w 0 i w0 N + = . x s b x 4 x 2 t 2
(23.73)
i = N0 , In the case where the beam is subjected to an uniform compression N x Equation (23.73) takes the form:
4w 0 x 4 4w 0 x 4
b 2w 0 2w 0 + + s N0 = 0, Ex I x 2 t 2 12 2w 0 2w 0 + N 0 = 0. s x 2 t 2 E x h3
(23.74)
or on introducing Expression (20.12) for the moment I : + (23.75)
The static buckling equation is deduced from these equations and may be written in one of the forms: d 4w 0 d x4 d 4w 0 d x4 b d 2w 0 + = 0, N0 Ex I d x2 + 12 E x h3 N0 d 2w 0 d x2 = 0.
(23.76)
These equations has a form identical to Equation (23.44) obtained in the case of buckling under cylindrical bending. The results obtained can thus be transposed to the present analysis. Equation (23.76) can be put in the form:
d 4w 0 d x4
+k
2
d 2w 0 d x2
=0,
(23.77)
with
23.3 Buckling of Beams
499
k2 =
bN 0 12 N 0 . = E x I E x h3
(23.78)
The general solution of this equation is of the form:

w 0 ( x) = A cos kx + B sin kx + Cx + D .
(23.79)
The coefficients A, B, C and D depend upon the conditions imposed at the ends of the beam x = 0 and x = L, where L is the length of the beam.
23.3.2 Simply Supported Beam

In the case of simple supports at the ends of the beam (Figure 23.7), the boundary conditions are at the ends x = 0 and x = L :
w 0 = 0,
M x = 0 or by (23.72)
d 2w 0 d x2
= 0.
(23.80)
Taking account of (23.79), the conditions at the end x = 0 lead to:

A + D = 0, A = 0.
Hence:
A= D = 0.
B sin kL + CL = 0, B sin kL = 0.
(23.81)
The conditions at the end x = L are then written as: (23.82)
In order to have a solution different from zero, it is necessary that:

kL = m .
(23.83)
It results that the coefficient B is arbitrary and that C is zero. The solution may
N0
N cr
FIGURE 23.7. Simply supported beam.
500
then be expressed in the form:

w 0 = Bm sin m
x . L
(23.84)
Relation (23.83) leads to the compression resultant:

N0 = m 2 2 E x h3 12 L2
(23.85)
The critical buckling load thus corresponds to the fundamental mode (m = 1). So:
N cr =
2 E x h3
12 L2
2 1
L2 D11
(23.86)
where D11 was expressed in (20.4). Comparison with Relation (23.50) in the case of a symmetric laminate shows that the difference between cylindrical bending . This difference is analoand beam bending results from the terms D11 and 1/ D11 gous to the difference between plane strain and plane stress states in the theory of elasticity. Cylindrical bending corresponds to a plane strain state and beam theory to a plane stress state. In general, the numerical results obtained are close in both cases.
23.3.3 Clamped Beam

In the case where the two ends of a beam are clamped, the conditions at x = 0 and x = L are: d w0 (23.87) w 0 = 0, = 0. dx Taking (23.79) into account, these conditions yield:
A + D = 0, kB + C = 0, A cos kL + B sin kL + CL + D = 0, Ak sin kL + kB cos kL + C = 0,
(23.88)
which leads to:
D = A, C = kB, A ( cos kL 1) + B ( sin kL kL ) = 0, A sin kL + B ( cos kL 1) = 0.
(23.89) (23.90)
A nonzero solution exists if the determinant of the matrix of the coefficients of A and B in Equations (23.90) is zero, which leads to the equation:
2 2 cos kL kL sin kL = 0 .
(23.91) (23.92)
This equation has as solution:
kL = 2m .
23.3 Buckling of Beams
501
N cr
FIGURE 23.8. Buckling of a beam with clamped ends.
It results that B = 0 and the solution can be expressed as:

x w 0 = D 1 cos 2m . L
(23.93)
Relation (23.92) leads to the expression for the compression resultant:

N0 = m 2 2 E x h3 3L2
(23.94)
The critical load corresponds to the fundamental mode (m = 1, Figure 23.8). Thus:
N cr =
2 E x h3
3L2
(23.95)
Comparing Equations (23.86) and (23.95) shows that clamping at the ends multiplies by 4 the critical buckling load obtained in the case of simply supported ends.
23.3.4 Other Support Conditions

Various support conditions can be studied in the same way. In the case of a beam clamped at one end and simply supported at the other, the critical buckling load is expressed as: 2 E x h3 . (23.96) N cr = 2.047 12 L2 The deformed shape is shown in Figure 23.9. In the case of a beam clamped at one end while the other end is free (Fgure 23.10), the critical load is given by:
N cr = 0.25
2 E x h3
12 L2
(23.97)
N cr
FIGURE. Buckling of a clamped-simply supported beam.
502
N cr
FIGURE 23.10. Buckling of a clamped-free beam.
23.3.5 Effect of Transverse Shear

In the case in which the transverse shear effect is taken into account, the buckling equations consist of Equation (20.86), and Equation (20.89) modified in such a way to take account of the transverse deformation. That is:
1 d 2w 0 d x + dx d x2 F55
D11 d x2
d 2w 0 N = 0, 0 d x2
(23.98) (23.99)
1 d 2 x
1 d w0 + = 0. x dx F55
These equations are similar to Equations (23.58) and (23.59) obtained in the case of cylindrical bending, the coefficients D11 and k55F55 being replaced by 1/ D11 and 1/ F55 , respectively. The results obtained in the case of cylindrical bending can thus be transferred to the case of buckling of beams. In particular, the critical buckling load of a simply supported beam can, from (23.65), be written as:
N cr = 1 1+ 2S , N cr
(23.100)
where
= N cr
2 1
L2 D11
(23.101)
and S' is a shear coefficient which can be expressed as a function of the coefficient S introduced in (20.101):
S =
F55 1 Ex h S h = = . 2 h D11 L 12 Gxz L 12
(23.102)
23.3.6 Buckling of a Sandwich Beam

i = N 0 , the In the case of a beam subjected to in-plane compression N x buckling equations of the sandwich beam are similar to Equations (23.67) and
23.4 Buckling of Orthotropic Plates under Biaxial Compression
503
(23.99) with:
1
F55
= hG13 .
(23.103)
The equations are:

d 2w 0 d x d 2w 0 hG13 + N = 0, 0 dx d x2 d x2
D11
(23.104) (23.105)
1 d 2 x
2
d w0 hG13 x + = 0. dx dx
These equations have the same form as Equations (23.67) and (23.68) obtained in the case of buckling under cylindrical bending. The critical buckling load of the beam is thus written in the same form:
N cr = 1 1+ 2S , N cr
(23.106)
2
with
= N cr
2 1
L2 D11
S S = , 12
E h S= x . G13 L
(23.107)
The variation of the critical load has the same form as that reported in Figure 23.6. In the case of a sandwich beam the core of which has a small shear modulus G13, the shear term 2 S can become large, strongly reducing the critical load.
23.4 BUCKLING OF ORTHOTROPIC PLATES UNDER BIAXIAL COMPRESSION 23.4.1 General Expressions
We consider (Figure 23.11) a rectangular plate simply supported along its four edges, made of an orthotropic laminate. This plate is subjected to a uniform compression on each edge, with respective resultants Nx and Ny, no transverse load being exerted (q = 0). The buckling equation is deduced from Equation (23.24), thus:
D11 4w 0 x 4 + 2 ( D12 + 2 D66 ) 4w 0 x 2y 2 + D22 4w 0 y 4 = Nx 2w 0 x 2 + Ny 2w 0 y 2
. (23.108)
The boundary conditions are written as: along the edges x = 0 and x = a :
w 0 = 0,
M x = D11 2w 0 x 2 D12 2w 0 y 2 = 0,
(23.109)
504
Ny
z
Nx
FIGURE 23.11. Rectangular plate subjected to bi-axial compression.

w 0 = 0,
M y = D12 2w 0 x 2 D22 2w 0 y 2 = 0.
(23.110)
The conditions are satisfied taking a deflection of the form:

w 0 ( x, y ) = Amn sin m
x y sin n . a b
(23.111)
Substituting this expression into Equation (23.108), we obtain:

4 2 2 2 4 4 2 Amn m D11 + 2m n R ( D12 + 2 D66 ) + n R D22
= Amn m 2 N x + n 2 R 2 N y a 2 , where R is the length-to-width ratio of the plate. A nonzero solution of the buckling problem is obtained when:
m2 N x + n2 R 2 N y =
(23.112)
2 4 m D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) + n 4 R 4 D22 . (23.113) 2

a
We consider the case of a uniform compression on each edge of the form: N x = N0 , N y = N 0 , (23.114)
where N0 is positive. Expression (23.113) leads to: N0 =

4 2 2 2 4 4 2 m D11 + 2m n R ( D12 + 2 D66 ) + n R D22 . 2 2 2 2 a m +n R
(23.115)
23.4 Buckling of Orthotropic Plates under Biaxial Compression
505
The critical buckling load corresponds to the values of m and n, which lead to the lowest values of N0. We study different types of loadings.
23.4.2 Uniaxial Compression

In the case of uniaxial compression in the x direction, we have = 0, and Expression (23.115) becomes:
N0 = m2a 2
2 4 2 2 2 4 4 m D11 + 2m n R ( D12 + 2 D66 ) + n R D22 .
(23.116)
For a given m, the smallest value of N0 is obtained for n = 1, which yields:

N0 = m2a 2
2 4 2 2 4 m D11 + 2m R ( D12 + 2 D66 ) + R D22 .
(23.117)
The expression of the critical buckling load depends upon the length-to-width ratio of the plate. In fact, we have: N 0 = N 0 (m + 1) N 0 (m) =
2
a m
2
2m + 1
2(
D 2 D22 11 m 2 ( m + 1) R 4 . (23.118) D22 m + 1)

2
The value of N 0 changes sign for:

D R = Rm = m ( m + 1) 11 D22
1/ 4
(23.119)
It results that: if R Rm : if R Rm :
N cr = N 0 (m), N cr = N 0 (m + 1).
1/ 4
(23.120) (23.121)
For the first values of m, this leads to: for R

D 2 11 D22 N cr = :
2 D + 2 R 2 ( D12 + 2 D66 ) + R 4 D22 , 2 11

a D 6 11 D22
1/ 4
(23.122)
for
D 2 11 D22
1/ 4
N cr =
4a
2 16 D11 + 8 R 2 ( D12 + 2 D66 ) + R 4 D22 , 2

R D 12 11 D22
1/ 4
(23.123)
for
D 6 11 D22
1/ 4
506
50 m=4
N cr
2 11
40
a2
30
m=3
Buckling load
20 m=2 10 m=1 0 0 1 2 3
1/ 4
Length-to-width ratio ( D22 D11 )
FIGURE 23.12. Variation of the critical buckling load as a function of the length-to-width ratio of a plate subjected to uniaxial compression.
(23.124) 9a The variation of the critical buckling load as a function of the length-to-width ratio of the plate is reported in Figure 23.12 for:
D11 = D22 , D12 + 2 D66 = 0, 7 D22 .
2 N cr = 2 81D11 + 18 R 2 ( D12 + 2 D66 ) + R 4 D22 .
(23.125)
For the values (23.119) of the length-to-width ratio, two buckling mode shapes leading to the same value of the critical load are possible:
w 0 = Am1 sin m
and
x y sin , a b
(23.126) (23.127)
x y w 0 = Am+1, 1 sin ( m + 1) sin . a b
23.4.3 Square Plate under Biaxial Compression

In the case of a square plate subjected to biaxial compression of the same value along the two edges, we have = 1 and R = 1. Expression (23.115) can
23.5 Buckling of Orthotropic Plates under Arbitrary Conditions
507
be written as: N0 =
2 4 2 2 4 m D11 + 2m n ( D12 + 2 D66 ) + n D22 . 2 2 2 (m +n ) a
(23.128)
This expression shows that, for D11 D22 , the smallest value of N0 is obtained for m = 1, that is: 2 D + 2 D66 D N0 = D22 11 + 2n 2 12 (23.129) + n4 . 2 2 (1 + n ) a D22 D22 For n = 1 : 2 1D D + 2 D66 N 0 (1) = 2 D22 11 + 2 12 (23.130) + 1 , 2 D22 D22 a and for n = 2 : 2 1D D + 2 D66 N 0 (2) = 2 D22 11 + 8 12 (23.131) + 16 . 5 D22 D22 a Comparison of Expressions (23.130) and (23.131) shows that, in the case where: D11 D + 2 D66 2 12 +9, D22 D22 the critical load corresponds to n = 1, which yields:
N cr = N 0 (1) ,
(23.132)
(23.133)
and the buckling mode is:

w 0 = A11 sin
x y sin . a b
(23.134)
In the case where:
D11 D + 2 D66 2 12 +9, D22 D22

N cr = N 0 (2) ,
(23.135)
the critical load corresponds to n = 2, hence: (23.136) (23.137) and the buckling mode is:
w 0 = A12 sin
x y sin 2 . a b
23.5 BUCKLING OF ORTHOTROPIC PLATES UNDER ARBITRARY CONDITIONS 23.5.1 General Expressions
We consider a rectangular plate made of an orthotropic laminate the material
508
directions of which coincide with the plate directions. This plate is subjected to uniform in-plane compressive forces (Nx along the edges x = 0 and x = a, Ny along the edges y = 0 and y = b) and uniform in-plane shear forces (Nxy along the four edges). The boundary conditions along the four edges are arbitrary for the moment. We search the solutions to the buckling problem using the Ritz method. In the absence of transverse loads, the association of Expressions (8.46), (16.38), (23.29) and (23.33) leads to the following expression for the total potential energy:
1 U= 2

x =0
b y =0
2 2 2w 0 2w 0 2w 0 2w 0 D11 + D22 + 2 D12 x2 x 2 y 2 y2
(23.138)
2 2 2 2w 0 w 0 + N w 0 + 2 N w 0 w 0 d x d y. N + 4 D66 + x y xy x x y y xy
The approximate solution is sought in the usual form of a double series in separate variables: w 0 ( x, y ) =
Amn X m ( x)Yn ( y) ,
m =1 n =1
(23.139)
where the functions Xm(x) and Yn(y) must constitute functional bases and are chosen so as to satisfy the boundary conditions along the four edges. The coefficients Amn are deduced from the stationarity conditions (8.62) which are: m = 1, 2, . . . , M , U (23.140) =0 n = 1, 2, . . . , N . Amn Proceeding as in Section 21.4 and taking into account Expression (21.107), we easily obtain the system of equations giving the coefficients Aij :
22 00 20 02 02 20 11 2 00 22 4 J nj + D12 ( I mi J nj + I mi J nj ) + 4 D66 I 11 mi J nj {D11Imi R + D22 I mi J nj R i =1 j =1 11 00 00 11 2 10 01 01 10 + a2 N x I mi J nj + N y I mi J nj R + N xy I mi J nj + I mi J nj R Aij = 0, M N
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
(23.141)
The system of equations (23.141) can also be written by expressing the product of integrals in the form (21.116). We obtain:
2200 2002 0220 2 0022 4 + D12 ( Cminj + Cminj ) + 4D66C1111 minj {D11Cminj R + D22Cminj R i =1 j =1 1100 0011 2 1001 0110 + a2 N xCminj + N y Cminj R + N xy Cminj + Cminj R Aij = 0, M N
) }
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
(23.142)
23.5 Buckling of Orthotropic Plates under Arbitrary Conditions
509
In the preceding equations, R is the length-to-width ratio of the plate. A nonzero solution of the system (23.141) or (23.142) is obtained in the case where the determinant of the matrix of the coefficients Aij vanishes. This condition allows to determine the combination of the lowest value of the resultants Nx, Ny and Nxy which makes the determinant vanish.
23.5.2 Clamped Orthotropic Plates Subjected to Uniform Shear

As an example of the general case treated in the preceding subsection, we consider the buckling problem of an orthotropic plate, clamped along its four edges and subjected to a uniform in-plane shear load along the four edges with resultant Nxy = S (Figure 23.13). The system of equations (23.142) reduces in this case to:
2200 2002 0220 1111 2 + D12 ( Cminj + Cminj ) + 4D66Cminj {D11Cminj R i =1 j =1 0022 4 1001 0110 R + a 2 S ( Cminj + D22Cminj + Cminj ) R} Aij = 0, M N
(23.143)
pour m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
The clamping conditions along the four edges have been written in (21.120) and (21.121). The approximate solution (23.139) is sought by using the functions Xm and Yn introduced in (21.131) and (21.132). These functions introduce the coefficients z a y
Nxy = S
x
FIGURE 23.13. Rectangular plate subjected to in-plane shear.
510
i and i (i = m, n) the values of which are reported in Table 21.3. Furthermore, the values of the integrals occurring in the system (23.143) have been determined in Subsection 21.5.3 (Expressions (21.139) to (21.142) and Tables B.1 to B.5 of Appendix B). When the integrals are evaluated, the system (23.143) can be solved numerically as an eigenvalue and eigenvector problem, for which the eigenvalues are the values S of the buckling modes and the eigenvectors determine the buckling mode shapes. The critical buckling load corresponds to the lowest of the values of S. For M = N = 1, the integrals multiplying S are zero, and the corresponding load is infinite. The first approximation is then obtained for M = N = 2. On introducing the numerical values of the integrals, Equations (23.143) are written:
2 4 2 500,56 D11 + 302.71( D12 + 2 D66 ) R + 500.56 D22 R A11 + 22.34a SRA22 = 0, 3803.5 D11 + 4241.2 ( D12 + 2 D66 ) R 2 + 3803.5 D22 R 4 A22 = 0, 22,34a 2 SRA11 + A12 = 0, A21 = 0. (23.144)
A nonzero solution is obtained by making the determinant zero, which leads to the critical buckling load:
2 2 4 Scr = 1.904 D11 + 3.808 D11D22 + 1.284 ( D11 + 2 D66 ) R
+ 3.274 ( D11 + D22 R
) ( D12 + 2 D66 ) R
2 8 + 1.904 D22 R
2 499.08 Ra (23.145) 106
1/ 2
The sign shows that the shear load can be positive or negative: there is no preferred direction for the critical shear. For the case of an orthotropic plate such that:
D11 = 10 D22 , D12 + 2 D66 = 1.2 D22 , D11 a2 a = b,
(23.146)
the approximation (23.145) leads, in the case of beam functions, to:

Scr = 74.297
(23.147)
TABLE 23.1. Critical buckling load of a square orthotropic plate subjected to in-plane shear.
Scr = k ( D11 / a 2 )
M=N k (beam functions) k (polynomials) 2 74.297 76.590 4 47.673 48.257 6 47.426 47.412 8 47.393 47.382 10 47.386 47.382
Exercises
511
FIGURE 23.14. Buckling mode shape of a square orthotropic plate subjected to in-plane shear.
The values obtained by using a larger number of terms in series (23.139) are reported in Table 23.1, in the case of approximations by beam functions and by polynomials. These values show that the convergence is slow and that the twoterm approximation (23.147) is very far from the actual value. In the case of an isotropic plate, Equation (23.145) reduces to:
Scr = 176 D a2 , D = D11 = D22 = D12 + 2 D66 ,
(23.148)
while the solution, for example [30], is:

Scr = 145 D a2
(23.149)
The two-term approximation leads to an error greater than 20%. Figure 23.14 gives the buckling mode shape of the orthotropic plate considered, obtained by using a large number of terms in the series (23.139). This figure shows the dissymetry of the deformation owed to the orthotropy of the plate.
EXERCISES
23.1 A beam is made of a symmetric laminate constituted of five layers. Layers 1, 3 and 5 are mat reinforced layers with the characteristics:
EL = ET = 8 GPa,
LT = 0.30,
GLT = 3 GPa.
Layers 1 and 5 have a thickness of 1 mm. Layer 3 has twice the thickness at 2 mm.
512
Layers 2 and 4 are unidirectional layers of thicknesses 1.5 mm and with the characteristics.
LT = 0.30,
GLT = 5 GPa.
Determine the critical buckling load of a beam of length L when the beam is simply supported or clamped at both ends. Do this question again when the preceding layers are modified in the following way. The laminate has three layers. Layers 1 and 3 are the previous mat reinforced layers of thickness 2mm. Layer 2 is a double layer of the previous unidirectional reinforcement with a thickness equal to 3 mm.
23.2 The five-layer symmetric laminate considered previously in Exercise 23.2 makes up half the skins of a sandwich beam the isotropic core of which is 30 mm thick and has the mechanical characteristics:
Ea = 70 MPa, Ga = 30 MPa.
Determine the critical buckling load of a beam of length L, clamped at both ends. Compare with the result obtained in Exercise 23.1.
23.3 A square plate made of the five-layer laminate considered in Exercise 23.1 is simply supported along its four edges. Express the critical buckling load (referred to the length a of the edge of the plate) and the buckling mode in the case of a uniaxial compression in the L, then in the T direction; and in the case of a biaxial compression. 23.4 Express the system (23.143) for M = N = 3. Derive the critical buckling shear load and the buckling mode of a square plate made of the five-layer symmetric laminate considered in Exercise 23.1. Do this question again in the case in which the plate is made of the sandwich material considered in Exercise 23.2.
CHAPTER 24
Vibrations of Laminate or Sandwich Beams and Plates
24.1 INTRODUCTION
When the time is taken into account, the general displacement field at point (x, y, z) at the instant t, for a first-order theory of the form (13.27), is written as:
u ( x, y, z, t ) = u0 ( x, y, t ) + z x ( x, y, t ), v ( x, y, z, t ) = v0 ( x, y, t ) + z y ( x, y, t ), w ( x, y, z, t ) = w 0 ( x, y, t ).
(24.1)
The functions u0, v0, w0, x and y are solutions of the fundamental relations (17.27) to (17.31). Neglecting the transverse shear effect, the functions x and y are expressed (14.2) as a function of w0, and the displacement field is completely determined by the knowledge of the functions u0, v0, w0. These functions are solutions of the fundamental relations (16.1) to (16.3), eventually taking account of equation (23.21) in the case of large transverse deformations under in-plane loading. The analysis of the vibrations of plates consists first in seeking the natural frequencies of vibrations. This investigation is carried out by expressing, for example in the absence of transverse shear, the displacement field in the form:
u0 ( x, y, t ) = u0 ( x, y ) eit , v0 ( x, y, t ) = v0 ( x, y ) eit , w 0 ( x, y, t ) = w 0 ( x, y ) eit , where is the angular frequency of the plate vibrations. The determination of the natural frequencies is next obtained by substituting these expressions into the fundamental relations. The natural frequencies can also be obtained by the Ritz (24.2)
514
Chapter 24 Vibrations of Laminate or Sandwich Beams and Plates
method (Section 8.4). The kinetic energy (16.42) is then expressed as follows:
1 Ec = 2 2
(u
s
2 0
2 2 ) dx dy , + v0 + w0
(24.3)
where the integral is extended to the plate dimensions.
24.2 CYLINDRICAL BENDING 24.2.1 Classical Laminate Theory

In the case of a laminate plate subjected to an initial compression load N0, no transverse load being applied (q = 0), the fundamental relations are given by Equations (19.2), (19.3) and (23.42):
A11 A16 2 u0 x 2 2u0 x 2 + A16 + A66 2v0 x 2 2v0 x 2 B11 B16 3w 0 x3 3w 0 x3 = s = s 2u0 t 2 2v0 t 2
, ,
(24.4) (24.5) (24.6)
D11
4w 0 x 4
B11
3u0 x3
B16
3v0 x3
+ N0
2w 0 x 2
+ s
2w 0 t 2
= 0.
So on taking account of (24.2), we have:

A11 A16 D11
d 4w 0 dx
4
d 2 u0 dx dx dx
2
+ A16 + A66 B16
d 2v0 dx dx dx
2
B11 B16 + N0
d 3w 0 dx
3
+ s 2u0 = 0 , + s 2v0 = 0 , s 2w 0 = 0 .
(24.7) (24.8) (24.9)
d 2u0
2
d 2v0
2
d3w 0 dx dx
3
B11
d 3u0
3
d 3v0
3
d 2w 0
2
In the case where the plate is simply supported along its edges x = 0 and x = a, the boundary conditions are satisfied by the displacements: x , a x v0 ( x) = Bm cos m , a x w 0 ( x) = Cm sin m . a u0 ( x) = Am cos m Substituting these expressions into Equations (24.7) to (24.9), we obtain:
(24.10)
24.2 Cylindrical Bending
515
2 2 m 2 2 m3 3 2 m s 2 A11 Am 2 A16 Bm + 3 B11Cm = 0, a a a
m3 3 a3
m 2 2 a2
m 2 2 m3 3 A16 Am + s 2 2 A66 Bm + 3 B16Cm = 0, (24.11) a a m 4 4 m 2 2 B16 Bm + 4 D11 2 N 0 s 2 Cm = 0. a a
B11 Am
m3 3 a3
As a function of the order of magnitude of the natural frequencies, it is possible to neglect the term s2 in the coefficients of Am and Bm in the two first equations. Equations (24.11) can then be written:
A11 Am A16 Bm +
A16 Am A66 Bm +
m3 3 a3 B11 Am m3 3 a3
m B11Cm = 0, (24.12) a
m B16Cm = 0, (24.13) a
m 4 4 m 2 2 B16 Bm + 4 D11 2 N 0 s 2 Cm = 0. (24.14) a a m B Cm , a A m C Cm , a A
Solving the first two equations leads to: Am = with

2 , A = A11 A66 A16 B = A66 B11 A16 B16 ,
Bm =
(24.15)
(24.16)
C = A11B16 A16 B11.
Substituting Am and Bm into the last equation (24.14), we obtain:
m 4 4 D m 2 2 2 N 0 s 2 Cm = 0 , 4 A a a with
D = D11 A B11B B16C .
(24.17) (24.18)
The coefficients A, B, C and D have already been introduced in the study of static cylindrical bending in (19.7) and (19.9). A nonzero solution of equation (24.17) is obtained in the case where the coefficient of Cm vanishes. Which leads to the expression for the natural frequencies of vibrations:
m m = a 1 m 2 2 D N0 , 2 s a A N 0 > 0.
(24.19)
516
If N0 = 0, Expression (24.19) of the natural frequencies can be written in the form:

1 H , m = m where the coefficient H was introduced in (23.52): H= B11B + B16C , AD11 (24.21) (24.20)
is the natural frequency of the bending vibrations in the case where there and m exists no in-plane flexural coupling (Bij = 0), expressed as:
= m
m 2 2 a
2
D11
(24.22)
The in-plane flexural coupling thus reduces the values of the natural frequencies of the bending vibrations. In the case of an initial in-plane compression of value N0, with N0 < Ncr where Ncr is the critical buckling load expressed in (23.48), the values (24.20) of the natural frequencies are reduced. The lowest frequency is written as:
1 =
1 2 D 2 N0 , s a A
0 N0
2 D
a2 A
(24.23)
In the case where an initial in-plane tensile load N0 is applied, the expression of the natural frequencies is:
m =
m a
1 m 2 2 D + N0 , 2 s a A
N 0 > 0.
(24.24)
The values of the natural frequencies of the bending vibrations are then increased. The fundamental frequency is expressed as:
1 =
1 2 D 2 + N0 . s a A
(24.25)
24.2.2 Taking account of the Transverse Shear

We now consider the effect of the transverse shear deformation on the vibration frequencies. In the case of orthotropic ( D16 = 0, D26 = 0) and symmetric (Bij = 0) laminates, in the absence of transverse loads, Equations (19.40) to (19.42) reduce to:
24.2 Cylindrical Bending
517
u0 = 0,
v0 = 0,
2w 0 2w 0 = k55 F55 x + , s x x 2 t 2 D11 2 x x 2 w 2 x k55 F55 x + 0 = I xy . x t 2
(24.26)
In the case of simple supports, the boundary conditions are given by Equations (19.47). The solutions x and w0 satisfying these boundary conditions and Equations (24.26), are, by extension of Expressions (19.49), of the form:
x a x w 0 = Cm eit sin m . a
x = Bm eit cos m ,
(24.27)
Substituting these expressions into Equations (24.26) yields:

m 2 2 m k55 F55 Bm + 2 k55 F55 s 2 Cm = 0, a a m 2 2 m 2 k55 F55Cm = 0. 2 D11 + k55 F55 I xy Bm + a a
(24.28)
A nonzero solution is obtained when the determinant of the preceding equations is zero. Hence the expression of the natural frequencies:
2 = m
1 2 s I xy
m 2 2 m 2 2 2 I xy + s k55 F55 + 2 s D11 , a a

2
(24.29)
with m 2 2 m 2 2 m 4 4 = 2 I xy + s k55 F55 + 2 s D11 4 4 s I xy k55 F55 D11 . (24.30) a a a In the case of a laminate made of layers of the same material, but having different orientation and thicknesses, the density of each layer is identical. From this it results that:
s = 0 h,
I xy = 0
h3 , 12
(24.31)
where 0 is the density of the material. The natural frequencies are then expressed as: 2 6 2 2 2 2 h 2 2 , a m k F m D m = + + (24.32) 55 55 11 12 0 a 2 h3 where is given by the expression :
518
m 4 4 h 4 h2 k55 F55 D11 . (24.33) = a 2 + m2 2 k55 F55 + m2 2 D11 12 3a 2 If we neglect the rotatory inertia terms (Ixy = 0), Equations (24.28) reduce to:
m 2 2 m k55 F55 Bm + 2 k55 F55 s 2 Cm = 0, a a m 2 2 m k55 F55Cm = 0. 2 D11 + k55 F55 Bm + a a
(24.34)
The natural frequencies can then be written in the form: m = m 1 1 + m 2 2 S , (24.35)
where S is the shear term introduced in (19.52):

S= D11 a k55 F55
2
(24.36)
is the natural frequency with shear deformation neglected, given by Exand m pression (24.22). The transverse shear deformation reduces the values of the natural frequencies. As in the case of static bending (Chapter 19), the influence of transverse shear on the values of the natural frequencies depends of the modulus ratio Q11 G13 (Relation (19.55)) and of the ratio a/h: span length between supports to thickness of the laminate. The variation of the fundamental frequency (m = 1) as a function of the ratio a/h reported in Figure 24.1, in the case of a [0/90/90/0] laminate the characteristics of which are given in (19.92).
24.2.3 Vibrations of Sandwich Plates

In the case of symmetric sandwich plates under cylindrical bending, in the absence of transverse loads, the fundamental relations (19.99) and (19.100) are written as: D11 2 x w 0 2 x hG I , + = xy 13 x x x 2 t 2 2w 0 2w 0 = . hG13 x + s x x 2 t 2
(24.37)
These equations have the same form as Equations (24.26). In the case of simple supports, the results are transposed from the results (24.29) to (24.36) by changing k55F55 into hG13. In particular, by neglecting the rotatory inertia terms, the natural frequencies are expressed by Relation (24.35):
24.3 Free Vibrations of Beams
519
1.2 classical theory
1 1 Fundamental frequency
1.0
0.8
k55 = 1
0.6
k55 =
1 3
transverse shear
0.4
0.2
10
15
Length-to-thickness ratio a h
FIGURE 24.1. Influence of the transverse shear deformation on the fundamental frequency of an orthotropic plate under cylindrical bending.
m = m with = m
1 1 + m 2 2 S D11 ,
(24.38)
m 2 2 a
2
(24.39)
where S is the shear term introduced in (23.70) :

S= D11 a hG13
2
(24.40)
24.3 FREE VIBRATIONS OF BEAMS 24.3.1 General Equation

In the case of a beam subjected to in-plane compression, the fundamental relation for the free vibrations is given by Expression (23.75):
520
4w 0 x 4
12 2w 0 2w 0 N + 0 = 0. s E x h3 x 2 t 2
(24.41)
The equation giving the natural frequencies is obtained by writing w0 in the usual form:
w 0 ( x, t ) = w 0 ( x)eit .
(24.42)
Substituting this expression into (24.41), we obtain: d 4w 0 d x4 + 12 d 2w 0 2 w N =0. 0 0 s d x2 E x h3 (24.43)
24.3.2 Simply Supported Beam

For a simply supported beam, the boundary conditions are (Relations (16.28)) : at the end x = 0: w 0 ( 0 ) = 0, M ( 0 ) = 0, (24.44) at the end x = L: w 0 ( L ) = 0, M ( L ) = 0, (24.45) the moment M being defined in (20.10). A solution of Equation (24.43), satisfying the boundary conditions is of the form:
w 0 ( x) = Cm sin m
x , a
m = 1, 2, . . ..
(24.46)
Substituting this expression into Equation (24.43) yields:

m 4 4 12 m 2 2 2 N + 4 Cm = 0 . s 0 Ex h3 L2 L
(24.47)
A nonzero solution is obtained only in the case where the coefficient of Cm vanishes. Hence the expression for the natural frequencies of vibrations:
m =
m L
1 m 2 2 E x h3 N0 , 2 s L 12
N 0 > 0.
(24.48)
If N0 = 0, the preceding expression reduces to:
m =
m 2 2 L2
E x h3 . 12 s
(24.49)
In the case of an initial in-plane compression N0, with N0 < Ncr where Ncr is the critical buckling load given in (23.86), the fundamental frequency is written:
521
1 =
1 2 E x h3 N0 , 2 s L 12
0 N0
2 E x h3
L2 12
(24.50)
In the case where an in-plane tensile load is applied of value N0, the expression of the natural frequencies is:
m m = L 1 m 2 2 E x h3 + N0 , 2 s L 12 N 0 > 0.
(24.51)
The values of the natural frequencies are increased by the presence of a tensile load.
24.3.3 Clamped Beam

In the case of a beam clamped at its two ends, the boundary conditions (16.30) are: at the end x = 0 : d w0 ( ) (24.52) 0 = 0, w 0 ( 0 ) = 0, dx at the end x = L : d w0 ( ) L = 0. (24.53) w 0 ( L ) = 0, dx These conditions are satisfied by writing the deflection in the form:
w 0 ( x) = Cm X m ( x),
m = 1, 2, . . . ,
(24.54)
where Xm(x) is the function introduced in (21.131), which here is expressed as:
X m ( x) = cos m x x x x cosh m m sin m sinh m . L L L L
(24.55)
The coefficients m are solutions of Equation (21.136) and m is expressed by Relation (21.137). The values of m and m are reported in Table 21.3 for m 8. On substituting Expression (24.54) into Equation (24.43), in the absence of an initial in-plane load (N0 = 0), we obtain:
4 m 12 2 4 Cm X m ( x ) = 0 . s E x h3 L
(24.56)
A nonzero solution for Cm is obtained only in the case where the coefficient of Cm vanishes. Whence the expression for the natural frequencies of the bending vibrations is:
m =
2 m
L2
E x h3 . 12 s
(24.57)
522
m =1
1 = 22.3730 2 = 61.6730 3 = 120.900
m=2
m=3
0 = 12
L
Ex h3 12 s
FIGURE 24.2. Bending vibration of a clamped beam.
The fundamental frequency is:
1 =
22,373 Ex h3 . 12 s L2
(24.58)
The deformed shapes derived from Relation (24.54) are reported in Figure 24.2 for m = 1, 2 and 3. The value of the amplitude Cm of the vibrations depends on the initial deformation imposed.
24.3.4 Beam Clamped at One End and Simply Supported at the Other
We consider the case of a beam clamped at the end x = 0 and simply supported at the end x = L. The boundary conditions are thus: at the end x = 0 :
w 0 ( 0 ) = 0,
at the end x = L :
w 0 ( L ) = 0,
d w0 ( ) 0 = 0, dx
M ( L ) = 0.
(24.59) (24.60)
The condition imposed for the moment is, by (20.10), equivalent to:
d 2w 0 ( ) L = 0, d x2
(24.61)
523
We express the deflection in the form introduced in (24.54):

w 0 ( x) = Cm X m ( x),
m = 1, 2, . . . ,
(24.62) (24.63)
with
X m ( x) = cos m x x x x cosh m m sin m sinh m . L L L L
These functions satisfy the conditions at the clamped end x = 0. It remains to satisfy the conditions: d2 X m ( ) X m ( L) = 0, L = 0. (24.64) dx Thus: cos m cosh m m ( sin m sinh m ) = 0, (24.65) cos m + cosh m m ( sin m + sinh m ) = 0.
A solution different from m = 0 is obtained when: cos m cosh m cos m + cosh m = , sin m sinh m sin m + sinh m or
tan m = tanh m .
(24.66)
The first eight solutions of this equation are reported in Table 24.1. The coefficient m is next determined by:
m =
cos m cosh m . sin m sinh m
(24.67)
The values of m are practically almost exactly equal to 1. It should be noted that for high enough values of m :
tanh m 1 ,
and Equation (24.66) reduces to:

tan m = 1 .
(24.68)
The solutions of this equation are:
m = ( m + 0, 25 ) .
(24.69)
The calculation of these approximate solutions shows that they are identical to the exact solutions of Equation (24.66).
TABLE 24.1. Values of the coefficient m of a simply supported-clamped beam function. m 1 3.927 2 7.069 3 4 5 6 7 8
10.210 13.352 16.493 19.635 22.776 25.918
524
i =1
1 = 15.4210 2 = 49.9710 3 = 104.240
i=2
i=3
0 = 12
L
Ex h3 12 s
FIGURE 24.3. Bending Vibration modes of a beam clamped at one end and simply supported at the other.
The natural frequencies of the bending vibrations are obtained by substituting (24.62) into Expression (24.43). In the absence of an initial in-plane load, the expression obtained is identical to Expression (24.57):
m =
2 m
L2
E x h3 . 12 s
(24.70)
The values of m are reported in Table 24.1. The fundamental frequency is for example, given by: 15, 421 Ex h3 . 1 = 12 s L2 (24.71)
The deformed shapes given by Relation (24.62) are reported in Figure 24.3 for m = 1, 2 and 3.
24.3.5 Beam Clamped at One End and Free at the other

In the case of a beam clamped at the end x = 0 and free at the end x = L, les the boundary conditions are: at the end x = 0 : d w0 ( ) 0 = 0, (24.72) w 0 ( 0 ) = 0, dx at the end x = L, by (16.32):
M x ( L ) = 0, Qx ( L ) = 0,
(24.73)
525
or by (20.10) and (20.18):

d 2w 0 ( ) L = 0, d x2 d3w 0 ( ) L = 0. d x3
(24.74)
The deflection is written again in the form (24.54). The function Xm(x) satisfies the clamping conditions at the end x = 0. It remains to satisfy:
d2 X m ( ) L = 0, d x2 d3 X m ( ) L = 0. d x3
(24.75)
That is:
cos m + cosh m m ( sin m + sinh m ) = 0, sin m sinh m + m ( cos m + cosh m ) = 0.
(24.76)
A nonzero solution for m is obtained when: cos m + cosh m sin m sinh m = , sin m + sinh m cos m + cosh m or
cos m cosh m = 1 .
(24.77)
The coefficient m is next determined by the expression:
m =
cos m + cosh m . sin m + sinh m
(24.78)
The first eight solutions of Equation (24.77) are reported in Table 24.2, with the corresponding values of m. For high enough values of m, approximate values can be expressed as: m = ( m 0,5 ) . (24.79) These values are also reported in Table 24.2 and show that they are practically the same as the solutions (24.77) for m 3. The natural frequencies are obtained by substituting Expression (24.54) for the deflection into Equation (24.43), which leads to Expression (24.57) for the natural frequencies. The fundamental frequency is written as:
1 =
3,516 Ex h3 . 12 s L2
(24.80)
TABLE 24.2. Coefficients of the clamped-free beam function.

m 1 1.875 0.734 1.571 2 4.694 1.018 4.712 3 7.855 0.999 7.854 4 10.996 1.000 10.996 5 14.137 1.000 14.137 6 17.279 1.000 17.279 7 20.420 1.000 20.420 8 23.562 1.000 23,562
m m
(m 0.5)
526
i =1
1 = 3.5160 2 = 22.0340 3 = 61.7010
i=2
i=3
0 = 12
L
Ex h3 12 s
FIGURE 24.4. Bending vibration modes of a beam clamped at one end and free at the other.
The deformed shapes of the bending vibrations given by Relation (24.54) with the values m and m of Table 24.2 are reported in Figure 24.4 for m = 1, 2 and 3.
24.3.6 Beam with Two Free Ends

In the case of a beam that is free at both ends, the boundary conditions are:
d2 X m ( ) d3 X m ( ) d2 X m ( ) d3 X m ( ) = = = 0 0, 0 0, L 0, L = 0. (24.81) d x2 d x3 d x2 d x3 The deflection is expressed again in the form (24.54), the first two conditions being satisfied by expressing the function Xm(x) in the form: x x x x + cosh m + m sin m + sinh m . L L L L The two conditions at the free end x = L are satisfied if: X m ( x) = cos m
(24.82)
cos m + cosh m + m ( sin m + sinh m ) = 0, sin m + sinh m + m ( cos m + cosh m ) = 0.
(24.83)
A nonzero solution for m is obtained when the determinant of these equations is zero. That is:
( cos m + cosh m )2 ( sinh 2 m + sin 2 m ) = 0 ,

cos m cosh m = 1 .
or
(24.84)
The coefficient m is next given by:
m =
sin m + sinh m . cos m cosh m
(24.85)
24.4 Vibrations of Simply Supported Rectangular Orthotropic Plates
527
TABLE 24.3. Values of the coefficients of the free-free beam function.
1 0
2 0
3 4.730
4 7.853
5 10.996
6 14.137 1.000
7 17.279 1.000
8 20.420 1.000
9 23.562 1,000
m m
0.9825 1.0008 1.000
The first root, zero, is a double root corresponding to the rigid motion of the beam, which can be expressed in the form:
w 0 ( x) = C1 X1 ( x) + C2 X 2 ( x) ,
(24.86) (24.87) (24.88)
with
X1 ( x ) = 1 , x X 2 ( x) = 3 1 2 . L
These functions correspond respectively to the rigid motions of translation and rotation of the beam. They are normalized according to Relation (21.139). The two roots 1 = 0 and 2 = 0 are associated to these functions. The other roots m of Equation (24.84) and the corresponding values of m are identical to those found in the case of two clamped ends (Relations (21.136) and (21.137)). The values of m and m are reported in Table 24.3, for m varying from 1 to 9. The first mode of the free bending vibrations is obtained for m = 3. The natural frequencies are identical to those of a beam with clamped ends. In contrast, the deformed shapes are different (Expression (24.82)). They are shown in Figure 24.5 for the first three modes.
24.4 VIBRATIONS OF SIMPLY SUPPORTED RECTANGULAR ORTHOTROPIC PLATES

In the case of an orthotropic laminate (a symmetric laminate for which D16 = D26 = 0), the fundamental relations (16.4) to (16.6) are written, taking into account the results established in Subsection 21.2.1 and in the absence of transverse loads (q = 0), in the form:
u0 = 0, D11 4w 0 x 4 + 2 ( D12 + 2 D66 ) v0 = 0, 4w 0 x 2y 2 + D22 4w 0 y 4 + s 2w 0 t 2 (24.89)
4w 4w = I xy 2 02 + 2 02 . x t y t
528
i =1
1 = 22.3730 2 = 61.6730 3 = 120.900
i=2
i=3
0 = 12
L
Ex h3 12 s
FIGURE 24.5. Bending Vibration modes of a beam with both ends free.
In the case where the rotatory inertia terms can be neglected (Ixy = 0), the last equation reduces to:
D11 4w 0 x 4 + 2 ( D12 + 2 D66 ) 4w 0 x 2y 2 + D22 4w 0 y 4 + s 2w 0 t 2 = 0.
(24.90)
The deflection expressed in the form:

w 0 ( x, y, t ) = w 0 ( x, y )eit ,
(24.91)
where is the angular frequency of the vibrations, leads, by substituting this expression into Equation (24.90), to:
D11 4w 0 x
4
+ 2 ( D12 + 2 D66 )
4w 0 x y
2 2
+ D22
4w 0 y
4
s 2w 0 = 0 .
(24.92)
In the case of simply supported edges, the boundary conditions are given by Relations (21.4) to (21.7), and w0(x, y) can be put in the form: w 0 ( x, y ) = Cmn sin m x y sin n , a b (24.93)
deduced from Expression (21.12) and satisfying the support conditions. Substituting this expression into Equation (24.92) yields: m 4 4 m 2 n 2 4 n 4 4 2 2 2 D D D D + + + ( ) 4 Cmn = 0 . (24.94) s 11 12 66 22 a 2b 2 b4 a
24.4 Vibrations of Simply Supported Rectangular Orthotropic Plates
529
A nonzero value of Cmn is obtained if the coefficient of Cmn is zero, whence the expression for the natural frequencies:
mn =
2
a2
1 4 m D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) + n 4 R 4 D22 , (24.95)

s
where R is the length-to-width ratio (a/b) of the plate. In the case of an isotropic plate (D11 = D22 = D12 + 2D66 = D), the expression for the natural frequencies is reduced to:
mn =
2
a
2
m 4 + 2m 2 n 2 R 2 + n 4 R 4 .
(24.96)
The deformed shape of the plate corresponding to the natural frequency mn is given by Expression (24.93). The fundamental frequency of an orthotropic plate corresponds to m = n = 1 and is given by:
11 =
2
a
2
1 D11 + 2 R 2 ( D12 + 2 D66 ) + R 4 D22 ,

s
(24.97)
and, in the case of an isotropic plate, it is written as:
11 =
2
a
2
(1 + R 2 ) .
(24.98)
The deformed shape of the fundamental mode is given in the two cases by: w 0 ( x, y ) = C11 sin x y sin . a b (24.99)
So as to appreciate the influence of the anisotropy, we compare the behaviour of a square plate made of an orthotropic material with characteristics:
D11 = 10 D22 , D12 + 2 D66 = D22 ,
(24.100)
with the behaviour of a plate made of an isotropic material. In the case of the isotropic material, the natural frequencies of vibrations (24.96) are written:
s a whereas for the orthotropic plate the natural frequencies are expressed as:
2
mn = kmn
kmn = m 2 + n 2 ,
(24.101)
mn = kmn
with
2
a
2
D22
(24.102)
kmn = 10m 4 + 2m 2 n 2 + n 4 .
(24.103)
530
TABLE 24.4. Frequencies and vibration modes of an isotropic square plate simply supported along its four edges.
mn = kmn
1st mode 2nd mode
2
a
2
s
4th mode
3rd mode
m n kmn
y nodal lines
1 1 2.0
y
1 2 5.0
y
2 1 5.0
y
2 2 8.0
y
1 3 10.0
y
3 1 10.0
TABLE 24.5. Frequencies and vibration modes of an orthotropic square plate simply supported along its four edges.
mn = kmn
1st mode 2nd mode 3rd mode
2
a2
D22
4th mode
5th mode
6th mode
m n kmn
y nodal lines
1 1 3.61
y
1 2 5.83
1 3 10.44
y y
2 1 13.0
2 2 14.42
y
1 4 17.26
y
24.5 Vibrations of Orthotropic Plates with Various Conditions along the Edges
531
The values of the natural frequencies and the corresponding modes of vibrations are reported in Table 24.4 for the isotropic plate and in Table 24.5 for the orthotropic plate. The results obtained show that there is no privileged direction in the case of an isotropic plate: the natural frequencies are the same for m = 1, n = 2 and m = 2, n = 1; as well as for m = 1, n = 3 and m = 3, n = 1, etc. In contrast, in the case of the orthotropic plate, for example, the second mode corresponds to m = 1, n = 2 (with kmn = 5.83), whereas m = 2, n = 1 corresponds to the fourth mode (with kmn = 13.0), etc.
24.5 VIBRATIONS OF ORTHOTROPIC PLATES WITH VARIOUS CONDITIONS ALONG THE EDGES 24.5.1 General Expressions
In the preceding Section, we obtained the exact solutions of Equation (24.90) in the case of a plate simply supported along its four edges. In the case of other support conditions, it is not possible to solve Equation (24.90) directly. The determination of the natural frequencies and the vibration modes then requires to use approximate methods. Hereafter, we develop the Ritz method (Section 8.4) for the analysis of the flexural vibrations of a rectangular plate. In the case of orthotropic laminates, the strain energy Ud is given by Expression (21.97), when the kinetic energy is written, by (16.42), on introducing w0 in the form (24.91), as follows:
Ec max = 1 2

x =0
b y =0
2 dx dy . s 2w 0
(24.104)
In the absence of transverse loads, the maximum energy function (Relation (8.65)) reduces to Ud max Ec max with:
U d max Ec max
1 = 2

x =0
b y =0
2 2 2 2 2w 0 2w 0 w w 0 0 D11 x 2 + 2 D12 x 2 y 2 + D22 y 2
2 2w 0 2 2 + 4 D66 w s 0 d x d y. xy
(24.105)
The approximate solution is sought in the usual form of a double series in separate variables: w 0 ( x, y ) =
Amn X m ( x) Yn ( y) ,
m =1 n =1
(24.106)
where the functions Xm(x) and Yn(y) have to satisfy the boundary conditions along
532
the edges x = 0, x = a and y = 0, y = b. The coefficients Amn are determined by the stationarity conditions (8.66): d max E c max ] = 0 [U Amn
m = 1, 2, . . . , M , n = 1, 2, . . . , N ,
(24.107)
d max E c max is the energy obtained by substituting Expression (24.106) where U for the deflection into Expressions (24.104) and (24.105). Taking account of Expression (21.107), the stationarity conditions (24.107) then lead to M N homogeneous equations:
22 00 20 02 02 20 11 11 2 J nj + D12 ( I mi J nj + I mi J nj ) + 4 D66 I mi J nj R {D11I mi i =1 j =1 00 22 4 00 00 + D22 I mi J nj R s a 4 2 I mi J nj Aij = 0, (24.108) M N
for m = 1, 2, . . . , M ,
n = 1, 2, . . . , N ,
pq rs where the integrals I mi were introduced in (21.108) and (21.109). and J nj
On expressing the product of integrals in the form (21.116), the system of equations (24.108) can be rewritten in the form of a dimensionless system as follows:
2200 2002 0220 2 + 12 ( Cminj + Cminj ) + 4 66C1111 minj {Cminj R
i =1 j =1
0022 4 0000 + 22Cminj R 2Cminj Aij = 0,
(24.109)
pour m = 1, 2, . . . , M ,
n = 1, 2, . . . , N ,
writing the flexural bending stiffnesses Dij as functions of D11:

D12 = 12 D11 , D66 = 66 D11, D22 = 22 D11,
(24.110)
and introducing the reduced frequency:
= a2
s
D11
(24.111)
Comparison of Equations (21.114), (23.141) and (24.108) shows the similarity between the equations obtained by the Ritz method in the case of bending, of buckling and of vibrations. The similarity results from the part of the expression for the strain energy Ud common to these equations. As the system of equations (24.108) or (24.109) in Aij is homogeneous, a nonzero solution is obtained when the determinant of the system vanishes. This condition leads to an equation the solutions of which are the natural frequencies mn of the flexural vibrations of the plate.
533
24.5.2 Rayleigh Approximation

The Rayleigh approximation consists of using for a given mode mn the dominant term of the series (24.106):
w mn ( x, y ) = Amn X m ( x) Yn ( y ) .
(24.112)
The natural frequency of the mode is then obtained by equating the maximum strain energy with the maximum kinetic energy associated to the maximum transverse displacement wmn. From (24.104), the maximum kinetic energy is written as:
00 00 2 c max = 1 s 2 ab I mm E J nn Amn , 2
(24.113)
and the maximum strain energy is deduced from Expression (21.97):

2 22 00 20 02 11 11 2 d max = 1 Amn U D11I mm J nn + 2 ( D12 I mm J nn + 2 D66 I mm J nn ) R 2 (24.114) 00 22 4 J nn R ab. + D22 I mm
The equality of the two Expressions (24.113) and (24.114) leads, with the notations already introduced, to: B 2 mn = mmnn , (24.115) 0000 Cmmnn where the coefficient Bmmnn is expressed as:
2200 2002 1111 0022 4 ) R 2 + 22Cmmnn Bmmnn = Cmmnn + 2 (12Cmmnn + 2 66Cmmnn R . (24.116)
In the case of transverse vibrations of orthotropic plates, the difference between the value of the natural frequency obtained by Rayleigh approximation and the value deduced from an approximation with a large number of terms (24.109) is low (less than a few percent) in the case of a plate having its edges clamped or simply supported. This difference increases when the geometric constraints imposed on the four edges decrease. Schematically, the change of a clamped edge or simply supported edge into a free edge increases the difference noticeably. The intersection of two free edges produces the highest differences.
24.5.3 Two-Term Approximation

In the case of a two-term, the transverse displacement is expressed, for example, as: w 0 ( x, y ) = A11 X1 ( x) Y1 ( y ) + A12 X1 ( x) Y2 ( y ) , (24.117) and the system of equations (24.109) reduces to a system of two equations :
534

0000 0000 ( B1111 2C1111 ) A11 + ( B1112 2C1112 ) A12 = 0, 0000 0000 ( B1112 2C1121 ) A11 + ( B1122 2C1122 ) A12 = 0,
(24.118)
with
2200 2002 1111 2 0022 4 B11ij = C11 ij + 2 12C11ij 2 66C11ij R + 22C11ij R = 0, i, j = 1, 2.
(24.119) The natural frequencies of modes 11 and 12 are obtained by making the determinant of the system (24.118) zero. That is:
0000 B1111 2C1111 det 0000 B1112 2C1121 0000 B1112 2C1112 =0. 0000 B1122 2C1122
(24.120)
The Rayleigh approximations of the two modes 11 et12 are obtained directly from the diagonal terms. Thus:
2 11 =
B1111
0000 C1111
et
2 12 =
B1122
0000 C1122
(24.121)
We recover the approximation given by Expression (24.115).
24.5.4 Orthotropic Plate with Simply Supported or Clamped Edges

As an application, we consider in this subsection the case of a rectangular plate clamped or simply supported along its edges. In the case of opposite edges being clamped, it is possible to use the beam functions introduced in (21.131) and (21.132): for clamped edges x = 0 and x = a : X m ( x) = cos m x x x x cosh m m sin m sinh m , a a a a (24.122)
for clamped edges y = 0 and y = b : Yn ( y ) = cos n y y y y cosh n n sin n sinh n , b b b b (24.123)
where m, n, m and n are given by Relations (21.136) and (21.137):

cos i cosh i = 1, cos i cosh i , sin i sinh i i = m, n. (24.124)
i =
535
The values of the coefficients i and i are given in Table 21.3. The values reported show that:
1 = 4, 730,
i = ( i + 0,5 )
i = 2, 3, . . ..
(24.125)
In the case of simply supported opposite edges, the functions used are the sine functions introduced in (24.93): for simply supported edges x = 0 and x = a : X m ( x) = sin m x , a (24.126)
for simply supported edges y = 0 and y = b : Yn ( y ) = sin n y . b (24.127)
In the case where one edge is clamped and the other opposite edge is free, the transverse displacement is expressed by the beam function introduced in (24.63): for the edge x = 0 clamped and the edge x = a simply supported: X m ( x) = cos m x x x x cosh m m sin m sinh m , a a a a (24.128)
for the edge y = 0 clamped and the edge y = b simply supported: Yn ( y ) = cos n y y y y cosh n n sin n sinh n , b b b b (24.129)
where the coefficients m, n, m and n are given by Relations (24.66) and (24.67): tan i = tanh i , cos i cosh i , sin i sinh i i = m, n. (24.130)
i =
The values of the coefficients i, which are reported in Table 24.1, can be expressed (24.69), as:
i = ( i + 0, 25 ) .
(24.131)
The natural frequencies and the corresponding modes are next determined by introducing the different functions (24.122) to (24.131) in the system of equations (24.109). For these equations, we have the relations:
0000 Cmnij = 1, 2002 0220 Cmnij = Cmnij = C1111 mnij .
(24.132)
536
The system of equations (24.109) can then be written in the form:

2200 1111 2 0022 4 2 Cminj + 2 (12 + 2 66 ) Cminj R + 22Cminj R Aij = 0, (24.133) i =1 j =1 M N
pour m = 1, 2, . . . , M , with
2200 22 00 22 Cminj = I mi J nj = I mi ,
n = 1, 2, . . . , N ,
0022 00 22 22 Cminj = I mi J nj = J nj . (24.134)
11 11 C1111 minj = I mi J nj ,
The values of these integrals are reported in Tables B.1 to B.5 of Appendix B, in the case where two opposite edges are clamped. In the other cases, these integrals have to be evaluated. Restricting to the Rayleigh approximation (24.115), the natural frequency of the mode mn can be expressed in the form:
mn =
1 a
2
D11
2200 1111 2 0022 4 + 2 (12 + 2 66 ) Cmmnn Cmmnn R + 22Cmmnn R . (24.135)
In the case of two opposite edges simply supported:

2200 Cmmnn = m 4 4 , 1111 Cmmnn = m 2 n 2 2 , 0022 Cmmnn = n 4 4 .
(24.136)
In the case of two clamped opposite edges, or of one edge clamped and the other simply supported:
2200 4 Cmmnn = m , 0022 4 Cmmnn = n ,
(24.137) (24.138)
and
11 11 C1111 mmnn = I mm J nn .
The values of the coefficients m and n are reported in Table 21.3 in the case of two opposite edges clamped, and in Table 24.1 in the case of one edge clamped 11 11 and J nn are and the other simply supported. The values of the integrals I mm reported in Table B.2 of Appendix B in the case of two opposite edges clamped. These values show that:
11 I ii = 12,30, 11 I ii i ( i 2 )
i = 2, 3, 4, . . ..
(24.139)
Lastly, the evaluation of these integrals in the case of one edge clamped and the other simply supported shows that:
11 I ii i ( i 1)
i = 1, 2, 3, . . ..
(24.140)
Finally Expression (24.135), associated with Relations (24.136) to (24.140), show that the Rayleigh approximation of the natural frequency of the mode mn can be written in the form:
mn =
1 a
2
D11
4 4 + 2 (12 + 2 66 ) R 2c2 + 22 R 4c3 c1 ,
(24.141)
537
TABLE 24.6. Coefficients introduced in the expression for the natural frequencies of the bending vibrations of an orthotropic rectangular plate (Clamped edges: E or Simply supported edges: S).
Boundary conditions y C C C y S
S S
m 1 1 2, 3, 4, . . . x 2, 3, 4, . . . 1
c1
c3
c2
4.730 4.730
( m + 0.5) ( m + 0.5)
4.730
( n + 0.5)
12.3 = 151.3 12.3c3 (c3 2) 12.3c1 ( c1 2) c1 (c1 2) c3 ( c3 2)
2, 3, 4, . . . 1 2, 3, 4, . . .
4.730
( n + 0.5)
1, 2, 3, . . . x
1, 2, 3, . . .
m n
2 2 4
S y S C C y S C S y C C S y S C S S C C S
1, 2, 3, . . . x
1, 2, 3, . . .
( m + 0.25)
( n + 0.25)
c1 (c1 2) c3 ( c3 2)
1 2, 3, 4, . . . x
1, 2, 3, . . . 1, 2, 3, . . .
4.730
( m + 0.5)
n n
12.3n
2 2
2 2
n c1 (c1 2)
1 2, 3, 4, . . . x
1, 2, 3, . . . 1, 2, 3, . . .
4.730
( m + 0.5)
( n + 0.25) ( n + 0.25)
12.3c3 (c3 2)
c1 (c1 2) c3 ( c3 2)
1, 2, 3, . . . x
1, 2, 3, . . .
( m + 0, 25)
n c1 (c1 2)
2 2
538
where the coefficients c1, c2 and c3 are reported in Table 24.6 for each combination of clamped and simple supported edges of the plate. In the case of an isotropic plate (D11 = D22 = D12 + 2D66 = D), the expression for the natural frequencies can be written in the form:
mn =
1 a
2
4 4 + 2 R 2c2 + R 4c3 c1 .
(24.142)
For an isotropic square plate clamped along its four edges, the value drawn from Table 24.6 lead to the following expression for the fundamental frequency:
11 =
36.1 a
2
(24.143)
Using a 64-term series (M = N = 8), the solution of the system (24.133) leads to the expression:
11 =
35.99 a2
(24.144)
The value deduced from the one-term approximation is thus very close to the exact value. In the case of a square orthotropic plate, clamped along its four edges, with the characteristics: D11 = 10 D22 , D12 + 2 D66 = 1, 2 D22 , (24.145) the values of the natural frequencies obtained by the one-term approximation (24.141) are compared in Table 24.7 with the values obtained by using a 64-term series. These results show that the values deduced from the one-term approximation are sufficiently precise.
TABLE 24.7. Natural frequencies of flexural vibrations of an orthotropic square plate clamped along its four edges.
mn = kmn
1 a
2
kmn m 1 1 1 2 2 1 2 2 n 1 2 3 1 2 4 3 4 approximation (24.141) 24.227 31.889 47.480 63.163 68.504 70.722 79.740 98.460 64-term series 24.213 31.861 47.436 63.116 68.428 70.645 79.676 98.369
24.6 Vibrations of Symmetric Laminate Plates
539
24.6 VIBRATIONS OF SYMMETRIC LAMINATE PLATES 24.6.1 General Expressions

The analysis of the free flexural vibrations of symmetric laminate plates can be implemented in the same way as in Section 24.5. In the present case, the strain energy to consider is that introduced in (22.2). From this it results that the system (24.108) or (24.109) is modified by introducing the terms in D16 and D26. Whence the system of M N homogeneous equations:
2200 2002 0220 1111 2 0022 4 + D12 ( Cminj + Cminj ) + 4D66Cminj {D11Cminj R + D22Cminj R
i =1 j =1 M N
2101 1012 0121 3 4 2 0000 + 2 D16 C1210 minj + Cminj R + 2 D26 Cminj + Cminj R s a Cminj Aij = 0,
pour m = 1, 2, . . . , M ,
n = 1, 2, . . . , N .
(24.146)
As in the case of orthotropic plates, it is possible to consider the Rayleigh approximation (Subsection 24.5.2) or the two-term approximation (Subsection 24.5.3), according to expressions analogous respectively to (24.115) and (24.120). However, in the present case the Rayleigh approximation differs notably from the value obtained with a large number of terms by the Ritz method. In fact, in the case of symmetric laminate, the one-term approximation of the transverse displacement does not describe the actual displacement correctly enough.
24.6.2 Symmetric Plates with Clamped or Free Edges

As an application of the preceding general formulation, we consider here the case of a rectangular plate made of a symmetric laminate, the edges of which are clamped or free. The case of clamped opposite edges has already be considered in Subsection 24.5.4 (Relations (24.122) to (24.125)). In the case where one edge is clamped and the other opposite edge is free, it is possible to express the transverse displacement by the beam function introduced in Subsection 24.3.5: for the edge x = 0 clamped and the edge x = a free: X m ( x) = cos m x x x x cosh m m sin m sinh m , a a a a y y y y cosh n n sin n sinh n , b b b b (24.147)
for the edge y = 0 clamped and the edge y = b free: Yn ( y ) = cos n (24.148)
540
where the coefficients m, n, m and n are given by Relations (24.77) and (24.78): cos i cosh i = 1, cos i + cosh i , sin i + sinh i i = m, n. (24.149)
i =
The values of i and i are reported in Table 24.2. In the case of free opposite edges, the transverse displacement is expressed by the beam functions introduced in (24.82), (24.87) and (24.88): for the free edges x = 0 and x = a : X1 ( x) = 1, x 3 1 2 , (24.150) a x x x x X m ( x) = cos m + cosh m + m sin m + sinh m , m 3, a a a a X 2 ( x) = for the free edges y = 0 and y = b : Y1 ( y ) = 1, y 3 1 2 , b y y y y Yn ( y ) = cos n + cosh n + n sin n + sinh n , n 3. b b b b Y2 ( y ) = (24.151)
The coefficients m, n, m and n are expressed by Relations (24.84) and (24.85):
cos i cosh i = 1, sin i + sinh i i = , cos i cosh i
i = m, n 3.
(24.152)
The values of i and i are reported in Table 24.3. It is important to note that if the beam functions (24.147) to (24.152) satisfy the boundary conditions (24.81) exactly at the free ends of a beam, they satisfy the boundary conditions only approximately in the case of free edges of a plate. In fact, in the case of one edge free in the direction parallel to y, for example, the boundary conditions (16.32) are written as:
M x = 0, M xy y + Qx = 0.
(24.153)
The transverse shear resultant Qx is expressed by the fourth plate equation des (13.56). Thus the boundary conditions are written as:
M x = 0, M xy M x +2 = 0. x y
(24.154)
24.6 Vibrations of Symmetric Laminate Plates
541
The expressions for the bending moment Mx and the twisting moment Mxy are deduced from the constitutive equation (14.29) for laminates. The boundary conditions along a free edge parallel to the y thus, finally, are written as:
D11 2w 0 x 2 + D12 2w 0 y 2 2w 0 + 2 D16 = 0 , (24.155) xy
D11
3w 0 x3
+ 4 D16
3w 0 x 2y
+ ( D12 + 4 D66 )
3w 0 xy 2
+ 2 D26
3w 0 y 3
= 0 . (24.156)
In the case of a free edge parallel to the x direction, these conditions are easily transposed by interchanging the respective roles of the variables x and y, and of the indices 1 and 2. The beam functions (24.147) and (24.150), in the x direction, satisfy the conditions (24.81) for the free ends of a beam, that is:
2w 0 x 2 = 0, 3w 0 x3 = 0.
(24.157)
From this it results that conditions (24.155) and (24.156) are only approximated by the beam functions. The approach by the Ritz method is less precise in the case of free edges. Considering the functions corresponding to the conditions imposed along the pq rs four edges of the plate, it is possible to evaluate the integrals I mi and J nj , and to establish the corresponding system (24.146) of homogenous equations. This system can be solved as an eigenvalue and eigenvector problem, where the eigenvalues are the natural frequencies of the plate vibrations and the eigenvectors determine the free modes of the vibrations. These calculations are greatly helped by the use of a general-purpose software package for scientific and engineer applications that integrates numerical analysis, matrix computation and graphics. As a numerical application, we consider the case of a plate made of an orthotropic laminate with flexural stiffnesses in its material directions:
0 0 D22 = 0.25 D11 , 0 0 D12 = 0.075 D11 , 0 0 D66 = 0.125 D11 .
(24.158)
The material directions are oriented at 30 to the geometric directions of the plate. The flexural stiffnesses with respect to the plate directions are then determined by applying to the coefficients (24.158) the transformation relations reported in Table 11.6. We obtain:
0 D11 = 0, 70 D11 , 0 D22 = 0,325D11 , 0 D12 = 0,1875D11 , 0 D26 = 0, 0974 D11 , 0 D16 = 0, 2273D11 , 0 D66 = 0, 2375D11 .
(24.159)
The values obtained for the natural frequencies of the first six modes are reported in Table 24.8, for the different combinations of clamped or free edges. The frequencies have been calculated by taking a 64-term as the displacement function. The deformed shapes of the modes are reported in Figures 24.6 and 24.7
542
TABLE 24.8. Natural frequencies of the flexural vibrations for the first six modes of a square plate made of a symmetric laminate, (clamped edges: C, free edges: F).
i =
ki a2
0 D11
for mode i
ki Boundary conditions mode 1 mode 2 mode 3 mode 4 CCCC FFFF CCFF CFCF CCCF CFFF 25.670 8.311 5.429 18.096 18.995 2.693 45.090 11.645 15.108 19.723 28.191 6.145 58.648 18.532 22.092 30.478 47.226 15.698 71.211 19.577 31.833 49.198 51.570 17.373
mode 5 mode 6 82.994 100.929 26.853 36.077 39.625 51.835 52.061 52.282 62.619 74.397 23.521 34.431
in the case of four clamped edges (Figure 24.6), and in the case of two adjacent edges clamped with the other two free (Figure 24.7).
24.7 VIBRATIONS OF NONSYMMETRIC LAMINATE PLATES 24.7.1 Plate Made of an Antisymmetric Cross-Ply Laminate
We consider the case of a rectangular plate of dimensions a and b, constituted of a [0/90]p cross-ply laminate. This laminate is characterized by: A16 = A26 = 0, A22 = A11, B12 = B16 = B26 = B66 = 0, B22 = B11, D16 = D26 = 0, (24.160) D22 = D11.
By introducing Expressions (24.2) for the displacements in Equations (16.1) to (16.3), in the absence of transverse loads (q = 0) and neglecting the rotatory inertia terms, we obtain: A11 2u0 x 2 + A66 2u0 y 2 + ( A12 + A66 ) 2v0 3w 0 B11 = 0, xy x3 (24.161)
( A12 + A66 )
2u0 2v 2v 3w 0 + A66 20 + A11 20 + B11 = 0, xy x y x3
3u0 3v0 4w 0 4w 0 4w 0 2 2 s 2w 0 = 0. D11 D D B + + + 3 ( ) 12 66 11 4 2 2 3 x 4 y x y y x
24.7 Vibrations of Nonsymmetric Laminate Plates
543
FIGURE 24.6. Free flexural modes of a square plate made of symmetric laminate, clamped along its four edges.
FIGURE 24.7 Free flexural modes of a square plate made of symmetric laminate, two adjacent edges of which are clamped and the other two are free.
544
In the case of hinged edges, free in the normal direction, the boundary conditions are written: along the edges x = 0 and x = a :
w 0 = 0, v0 = 0,
M x = B11
u0 2w 0 2w 0 D D11 = 0, 12 x x 2 y 2
u v 2w 0 N x = A11 0 + A12 0 B11 = 0, x y x 2 v0 2w 0 2w 0 D12 D = 0, 11 y x 2 y 2
(24.162)
w 0 = 0,
u0 = 0,
M y = B11
u v 2w 0 N y = A12 0 + A11 0 + B11 = 0. x y x 2

x y sin n , a b x y v0 = Bmn sin m cos n , a b x y w 0 = Cmn sin m sin n . a b u0 = Amn cos m
(24.163)
The boundary conditions are satisfied with displacements of the form:
(24.164)
Substituting these expressions into Equations (24.161), we obtain: a1 Amn + a2 Bmn + a3Cmn = 0, a2 Amn + a4 Bmn + a5Cmn = 0, s a2 2 a3 Amn + a5 Bmn + a6 2 Cmn = 0, with a1 = m 2 A11 + n 2 R 2 A66 , a2 = mnR ( A12 + A66 ) , a3 = m3 (24.165)
B11, (24.166)
a4 = m 2 A66 + n 2 R 2 A11, a5 = n3 R3 a6 = a2 a R= . b
B11,
2 ( 4 4 4 ) m + n R D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) ,
545
A nonzero solution is obtained when the determinant of the system (24.165) of homogenous equations vanishes. This condition leads to the following expression for the natural frequencies:
2 mn =
4 ( 4 4 4 ) m + n R D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) 4 s a
2 B11
( m 3 + n
4
4 4
R 2 ) ,
(24.167)
on setting:
1 = ( m 2 A11 + n 2 R 2 A66 )( m 2 A66 + n 2 R 2 A11 ) m 2 n 2 R 2 ( A12 + A66 ) ,

2
2 = m 4 ( A12 + A66 ) + m 2 n 2 R 2 A11 + n 4 R 4 A66 ,

2
(24.168)
3 = m 4 A66 + m 2 n 2 R 2 A11 + n 4 R 4 ( A12 + A66 ) .

When the in-plane flexural coupling is neglected ( B11 = 0 ), Expression (24.167) of the natural frequencies reduces to:
2 mn =
4 ( 4 4 4 ) m + n R D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) , 4 s a
(24.169)
which is the expression (24.95) of the natural frequencies of flexural vibrations of orthotropic plates simply supported along its four edges, and in the case of the plate is made of a material for which D22 = D11. In the case of orthotropic laminates, Expression (24.169) shows that the fundamental frequency corresponds to m = n = 1 . It is not the same in the case where a coupling exists. The number of the mode then corresponding to the fundamental frequency can not be deduced in the general case from Expression (24.167). It depends on the characteristics of the layers constituting the laminate. We examine the case of antisymmetric cross-ply laminates made of layers the characteristics of which are: EL = 20 ET , GLT = 0.5ET ,
LT = 0.25.
(24.170)
The values of the stiffness are determined from Relations (22.40) to (22.46). The variation of the fundamental frequency as a function of the length-to-width ratio (a/b) of the plate is reported in Figure 24.8 in the case of [0/90], [0/90]2, [0/90]3 cross-ply laminates, and for an orthotropic laminate ( B11 = 0 ). The fundamental frequencies correspond in all cases to m = n = 1 . We observe that the inplane flexural coupling reduces the values of the natural frequencies. Also, the results obtained show that the fundamental frequencies rapidly tend to the solution (24.169) of the orthotropic laminate when the number of layers is increased.
546
40
ET h3
[ 0 / 90]3
30
fundamental frequency 11a 2
[ 0 / 90]2
20
orthotropic laminate (B11 = 0)
10
[ 0 / 90 ]
0.5
1.5
Length-to-width ratio a b
FIGURE 24.8. Variation of the fundamental frequency of a rectangular plate, made of an antisymmetric cross-ply laminate, as a function of the length-to-width ratio of the plate.
24.7.2 Plate Made of an Angle-Ply Laminate

In this Subsection, we consider the case of a rectangular plate, made of a [ ]n angle-ply laminate. This laminate is characterized by: A16 = A26 = 0, B11 = B12 = B22 = B66 = 0, D16 = D26 = 0. (24.171)
The introduction of Expressions (24.2) for the displacements into Equations (16.1) (16.3) leads, in the absence of transverse loads (q = 0) and neglecting the inplane inertia terms, to:
A11 2u0 x 2 + A66 2u0 y 2 + ( A12 + A66 ) 2v0 3w 3w 0 3B16 2 0 B26 = 0, xy x y y 3
( A12 + A66 )
D11 4w 0 x 4
2u0 2v 2v 3w 0 3w 0 + A66 20 + A22 20 B16 3 B = 0, 26 xy x y x3 xy 2 + 2 ( D12 + 2 D66 ) 4w 0 x 2y 2 + D22 4w 0 y 4
(24.172)
3u0 3v0 3u0 3v0 B B16 + + s 2w 0 = 0. 3 3 2 x 2y x3 26 y 3 xy
547
In the case of hinged edges free in the tangential directions, the boundary conditions are written as: along the edges x = 0 and x = a : w 0 = 0, u0 = 0, 2w 0 2w 0 u0 v0 M x = B16 + D11 D12 = 0, y x 2 y 2 x v w0 w0 u N xy = A66 0 + 0 B16 B26 = 0, 2 x x y 2 y
2 2
(24.173)
along the edges y = 0 and y = b : w 0 = 0, v0 = 0, v 2w 0 2w 0 u M y = B26 0 + 0 D12 D = 0, 22 x x 2 y 2 y N xy v 2w 0 2w 0 u B = A66 0 + 0 B16 = 0. 26 x x 2 y 2 y x y cos n , a b x y v0 = Bmn cos m sin n , a b x y w 0 = Cmn sin m sin n . a b u0 = Amn sin m Substituting these expressions into Equations (24.172), we obtain: Cmn = 0, a1 Amn + a2 Bmn + a3 Bmn + a5 Cmn = 0, a2 Amn + a4 s a2 2 a3 Amn + a5 Bmn + a6 2 Cmn = 0. The system obtained has the same form as the system (24.165), with:
= a3
(24.174)
These boundary conditions are satisfied by considering displacements of the form:
(24.175)
(24.176)
a
2
nR ( 3m 2 B16 + n 2 R 2 B26 ) , m ( m 2 B16 + 3n 2 R 2 B26 ) ,
= m A66 + n 2 R 2 A22 , a4 = a5 = a6
a
2
(24.177)
4 m D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) + n 4 R 4 D22 . 2

a
The expression for the natural frequencies can then be written in a form analogous to Relation (24.167):
548
2 mn =
4 4 m D11 + 2m 2 n 2 R 2 ( D12 + 2 D66 ) + n 4 R 4 D22 4 s a
1 m ( m 2 B16 + 3n 2 R 2 B26 ) 2 + nR ( 3m 2 B16 + n 2 R 2 B26 ) 3 , 1 (24.178)
on setting: = ( m 2 A11 + n 2 R 2 A66 )( m 2 A66 + n 2 R 2 A22 ) m 2 n 2 R 2 ( A12 + A66 ) , 1

2
2 = ( m 2 A11 + n 2 R 2 A66 )( m 2 B16 + 3n 2 R 2 B26 )
3 = ( m 2 A66 + n 2 R 2 A22 )( 3m 2 B16 + n 2 R 2 B26 )
n 2 R 2 ( A12 + A66 ) ( 3m 2 B16 + n 2 R 2 B26 ) ,
(24.179)
n 2 R 2 ( A12 + A66 ) ( m 2 B16 + 3n 2 R 2 B26 ) .
When the in-plane flexural coupling is neglected ( B16 = B26 = 0 ), Expression (24.178) of the natural frequencies of the vibrations reduces to Expression (24.95) obtained in the case of orthotropic plates with simply supported edges. In the case where a coupling exists, the number of the fundamental mode depends on the characteristics of the layers constituting the laminate. We consider the case of a balanced angle-ply laminate, made of layers the characteristics of which are given in (24.170). The variation of the natural frequency (corresponding in this case to m = n = 1 ) is reported in Figure 24.9 for a square plate, in the case of [+ / ], [+ / ]2, [+ / ]3 laminates, and for an orthotropic laminate ( B16 = B26 = 0 ). The results obtained show that the values of the fundamental frequency rapidly tend to the solution (24.95) of the orthotropic laminate, when the number of layers increases.
EXERCISES
24.1 A beam is made of the five-layer symmetric laminate of Exercise 23.1. Plot the values of the frequencies of the first four modes as functions of the length L of the beam: in the case where the beam has its ends simply supported ; next, in the case where the ends are clamped ; and last, in the case of a clamped-free beam. 24.2 Do Exercise 24.1 in the case where the beam is made of the sandwich material considered in Exercise 23.2.
Exercises
549
20
ET h3
18
fundamental frequency 11a 2
orthotropic laminate (B16 = B26 = 0)
16
[ ]3 [ ]2
[ ]
14
12
10
10
15
20
25
30
35
40
45
Orientation ( )
FIGURE 24.9. Variation of the fundamental frequency of the vibrations of a square plate, made of an angle-ply laminate, as a function of the orientation of the layers.
24.3 We consider a plate made of an orthotropic symmetric sandwich material. Derive the equations for the flexural vibrations in the case where the rotatory inertia terms can be neglected. In the case of simple supported edges, the vibration modes are sought in the form: x y x ( x, y ) = Amn cos m sin n , a b x y y ( x, y ) = Bmn sin m cos n , a b x y w 0 ( x, y ) = Cmn sin m sin n . a b
Show that these functions satisfy the simple support conditions along the four edges. Derive the system of equations for the natural frequencies and modes. Deduce from them the expression for the natural frequencies.
24.4 Express the system (24.109) for M = N = 3 in the case of a clamped rectangular plate, using the beam functions. Starting from the system of equations so obtained, calculate the reduce natural frequencies and the corresponding modes of vibrations, for a plate of length twice the width and with the characteristics of the plate:
D12 = 0.08D11 ,
D66 = 0.12 D11 ,
D22 = 0.5 D11.
550
24.5 We consider an orthotropic rectangular plate clamped along two adjacent edges and free on the other two. In order to solve the vibration problem, we consider the beam functions (24.147) and (24.148). Do exercise 24.4 again for the present case. 24.6 We study the vibrations of the plate considered in Exercise 22.1. From the system obtained for the plate vibrations, calculate the natural frequencies (the reduced frequencies) and the corresponding modes of vibrations. 24.7 Do Exercise 24.6 again for the case in which the plate has two adjacent edges clamped and the other two free. The problem will be solved by using the beam functions (24.147) and (24.148).
CHAPTER 25
Effects of Expansional Strain on the Mechanical Behaviour of Laminates
25.1 INTRODUCTION
In general, the properties of composite materials are affected by the conditions of the environment to which they are subjected. Among the factors related to the environment, those which induce variations in strain in the absence of any external mechanical loading are of particular interest. In the case of composite material structures, these phenomena are the consequence of a variation in temperature, of absorption of swelling agents such as water vapour by the polymer matrix, of expansion of gas absorbed by the matrix, etc. These phenomena induce strains and stresses that can significantly modify the mechanical behaviour of composite material structures: stiffness, buckling, vibration frequencies, etc. In this chapter, we consider how the laminate equations are modified when these expansion phenomena are taken into account, as well as the consequences in the mechanical behaviour of plates made of laminates.
25.2 ELASTICITY RELATIONS INCLUDING EXPANSIONAL STRAIN EFFECTS 25.2.1 Elasticity Relations in Material Directions
The study of the mechanical behaviour of laminates has, up here, been carried out by considering that the material was referred to a reference temperature for which the strain field and the stress field in the material have been considered to be zero in the absence of any external mechanical loading. In practice, structures are submitted to variations in temperature during their manufacture as well as during their use. The first effect of a variation in temperature is to modify the stiffness and the fracture properties of the materials. Moreover, the variation in temperature produces a thermal expansion (extension or contraction) of the
552
Chapter 25. Effects of Expansional Strain on the Mechanical Behaviour of Laminates
materials. The thermal expansion phenomena can be described by introducing the thermal strains at point (x, y, z) and at instant t in the form:
i*(thermal) = i T ( x, y, z , t ),
i = 1, 2, . . . , 6,
(25.1)
where i are the coefficients of thermal expansion and T is the variation of temperature from a reference temperature for which the thermal strains are considered to be equal to zero. The distribution of temperatures within the structure and in the course of time is determined from heat transfer processes. The phenomena of expansion by absorption of humidity or gas lead to some effects analogous to thermal effects. The strains which result can be expressed in the form:
i(swelling) = i C ( x, y, z , t ),
i = 1, 2, . . . , 6,
(25.2)
where i are the swelling coefficients (for example the coefficients of hygrometric expansion), and C is the variation of the concentration of the swelling agent from a state in which the swelling strains are zero. The distribution of concentrations of a swelling agent is determined by physico-chimical concepts such as Ficks law [31]. So as to include the effects of the expansion phenomena, the law of elasticity (7.3), written in a reference state in which the strains induced by the expansion phenomena are zero, must be modified and written in the form:
i =
Sij j + i ,
j =1
i = 1, 2, . . . , 6,
(25.3)
where Sij are the compliance stiffnesses and i are the expansional strains induced by the thermal effects, by the swelling agents, etc.:
i = i(thermal) + i(swelling) + . . . .
The inverted form of the elasticity relation (25.3) is written as:
(25.4)
i =
Cij ( j j ),
j =1
i = 1, 2, . . . , 6,
(25.5)
where Cij are the stiffness constants. In practice, the thermal and swelling phenomena produce only extensions or contractions (called by the general term of expansion), not affecting the shear strains. In this case, the elasticity relations can be rewritten in the form:
i = i =
Sij j + i ,
j =1 6
i = 1, 2, 3,
(25.6)
i = 4, 5, 6,
Sij j ,
j =1
25.2 Elasticity Relations Including Expansional Strain Effects
553
and
i = i =
j =1 6 j =1
Cij ( j j)+
Cij j ,
j =4
i = 1, 2, 3,
(25.7)
i = 4, 5, 6.
Cij j ,
In the case of orthotropic materials, the elasticity relation (25.5), referred to the directions of the material is written as:
1 C11 C12 2 C11 C22 3 C13 C23 = 0 4 0 5 0 0 0 6 0 C13 C23 C33 0 0 0 0 0 0 C44 0 0 0 0 0 0 C55 0 0 1 1 2 2 0 0 3 3 . 0 4 4 0 5 5 C66 6 6
(25.8)
Thus, in practice the elasticity relation is:

1 C11 C12 2 C11 C22 C C23 3 = 13 0 4 0 0 5 0 0 0 6 C13 = C12 , C13 C23 C33 0 0 0 0 0 0 C44 0 0 0 0 0 0 C55 0 0 1 1T 1C 0 2 2 T 2 C 0 3 3T 3C , (25.9) 0 4 0 5 6 C66
with, in the case of a unidirectional material:

C33 = C22 , C44 =
1 2
( C22 C23 ) ,
C55 = C66 .
(25.10)
For a plane stress state (Section 11.3), Relation (25.8) reduces to:
1 Q11 Q12 2 = Q12 Q22 0 6 0
0 1 1 , 0 2 2 Q66 6 6
(25.11)
on introducing the reduced stiffness constants (11.47) of the material.
25.2.2 Off-Axis Elasticity Relations

When the material directions make an angle (Figure 11.1) with the reference directions (x, y ,z), the elasticity relation referred to these directions can be written
554
as an extension of Relation (11.3) in the form:
C12 C13 0 xx C11 C22 C23 0 yy C11 C C C 0 23 33 zz = 13 yz 0 0 0 C44 0 0 C45 xz 0 C C C 0 26 36 xy 16
0 0 0 C45 C55 0
xx * C16 xx yy * C26 yy zz * C36 zz , 0 yz * yz 0 xz * xz * C66 xy xy
(25.12)
* * * * * where * xx , yy , zz , yz , xz , xy are the expansional strains induced by the
expansion phenomena, referred to the reference directions (x, y). The expressions for the stiffness constants are those expressed in Table 11.3 as functions of the stiffness constants in the material directions. The relations giving the strains * * * * xx , yy , etc., as functions of the expansional strains 11 , 22 , etc., expressed in the material directions, are deduced from Relations (6.42) and (6.44), by noting that the transformation (1, 2, 3) (x, y, z) is obtained by a rotation through an angle . We have, for example :
* * 11 xx * * yy 22 * * zz = T 1 33 , * * 23 yz * * xz 13 * * 12 xy
(25.13)
where the transformation matrix T1 is expressed in (6.45). In the case of orthotropic materials, the elasticity relations referred to the material directions are described by Relations (25.6) and (25.7). In the material directions, the expansional shear strains are zero, hence:
* = * = * = 0. 23 13 12
(25.14)
Considering Relation (25.13), the expansional strains in the (x, y, z) system thus reduce to:
* cos 2 xx * yy sin 2 * = 1 zz * xy 2sin cos sin 2 cos 2 0 2sin cos sin cos * 11 sin cos * 22 . 0 * 33 cos 2 sin 2
(25.15)
It results that the elasticity relation (25.12), expressed in the (x, y, z) system, then reduces to:
25.3 Governing Equations of a Laminate
555
C12 C13 0 xx C11 C22 C23 0 yy C11 C C C 0 23 33 zz = 13 yz 0 0 0 C44 0 0 C45 xz 0 C C 0 26 36 xy C16
0 0 0 C45 C55 0
xx * C16 xx yy * C26 yy zz * C36 zz . 0 yz 0 xz xy * C66 xy
(25.16)
For a plane stress state, * zz = 0 , and Relation (25.15) takes the form:
* xx cos 2 * 2 yy = sin * xy 2sin cos
* 11 cos 2 * . 22 2sin cos sin 2
(25.17)
The elasticity relation, taking account of (11.43), is then written as:

Q12 xx Q11 Q22 yy = Q12 Q Q 26 xy 16 xx Q11 Q12 Q16 yy Q12 Q22 Q26 Q Q66 26 xy Q16 * Q16 xx * yy , Q26 * Q66 xy
(25.18)
* * where the strains * xx , yy , xy are expressed by Relation (25.17) as functions of * , * , referred to the material directions. The stiffness expansional strains 11 22 are expressed in Table 11.6. constants Qij
25.3 GOVERNING EQUATIONS OF A LAMINATE
25.3.1 Constitutive Equation

In the framework of the classical laminate theory, Relation (14.20) expressing the stresses in the layer k is replaced, on taking account of Expression (25.18), by the relation:
Q12 xx Q11 Q22 yy = Q12 Q 26 xy k Q16 Q16 Q26 Q66 k
0 + z Q xx x 11 Q12 0 Q22 yy + z y Q12 0 Q 26 xy + z xy Q16
Q16 Q26 Q66 k
* xx * yy . (25.19) * xy
The constitutive equation is next obtained by combining the preceding expression with the relations of definitions (13.17) and (13.19) of resultants and moments. We obtain:
556
N x A11 N y A12 N xy = A16 M x B 11 M y B12 M B16 xy
A12 A22 A26 B12 B22 B26
A16 A26 A66 B16 B26 B66
B11 B12 B16 D11 D12 D16
B12 B22 B26 D12 D22 D26
0 * B16 xx Nx 0 * B26 yy N y 0 * B66 xy N xy , * D16 x M x * D26 y My * D66 xy M xy
(25.20)
where the coefficients Aij, Bij, Dij are the stiffness constants of the laminate given by Relations (14.31) to (14.33), and by introducing the resultants and moments owed to the expansion processes defined as:
*, M *) = (Nx x
* * * xx + Q12 yy + Q16 xy )k (1, z ) d z , h (Q11

k =1 n
k 1
hk
* (N* y, M y) =
* * * xx + Q22 yy + Q26 xy )k (1, z ) d z , h (Q12

k =1 n
k 1
hk
(25.21)
* , M* ) = ( N xy xy
* * * xx + Q26 yy + Q66 xy )k (1, z ) d z. h (Q16

k =1
k 1
hk
* * The expansional strains ( * xx , yy , xy ) k in each layer are expressed as functions * , * ) , referred to the material directions of the of the expansional strains (11 22 k * , * ) themselves layer, by Relation (25.17). The expansional strains are (11 22 k evaluated by relations of the types (25.1) and (25.2).
The constitutive equation (25.20), taking account of the expansion phenomena, differs from the constitutive equation (14.29) of the initial classical theory by the addition of the resultants and moments induced by thermal phenomena, by the absorption of swelling agents, etc. The expansional stresses (thermal, hygrometric, etc.) expressed by Relation (25.19) are induced when the conditions of temperature, hygrometry, etc., of the laminate differ from the state where the laminate is free of all hygrothermal constraints. In fact, these constraints are not induced by the only expansional effects (extension or contraction) of the laminate, but simultaneously result from the expansional processes and the fact that the laminate and layers are not free to extend or contract. Indeed, no force or moment is induced in the laminate by the expansional processes, when the laminate is completely free to expand, bend and twist. However, each layer of the laminate influences the extension or contraction of the neighbouring layers, because of different mechanical and expansional properties. The layers are then not free to deform. The hygrothermic stresses in each layer thus result from the constraints imposed to their deformation by the neighbouring layers.
557
The thermal constraints are practically inevitable as a result of the elaboration of the laminates. The stresses, called residual stresses or curing stresses, are induced by cooling the laminates after curing at high temperatures. In some cases, such residual stresses can be sufficiently high to modify the fracture characteristics of the laminates. It is then necessary to take them into consideration in the design of laminate structures. In practice, the matrix has coefficient of thermal expansion greater than that of the fibre, inducing a radial compression of the fibres at the fibre-matrix interface. This radial compression allows a transfer of the loads from the matrix to the fibres by shear effect, even in the absence of good fibre-matrix bonding.
25.3.2 Examples
25.3.2.1 Calculation of Thermal Stresses
We consider the case of a symmetric cross-ply laminate (Figure 25.1), made of three unidirectional layers 1 mm thick, with mechanical characteristics:
LT = 0.31,
GLT = 4.5 GPa,
(25.22)
and with the thermal expansion coefficients:
L = 5 106 /C,
T = 20 106 /C .
(25.23)
The curing process of the laminate was carried out at a temperature of 120 C. We need to determine the residual stresses at the working temperature of 20 C. Referred to the material directions of the layers, the reduced stiffness constants of the layers are (11.52):
Q11 = 45.982 GPa, Q22 = 10.218 GPa,
Q11 Q12 Q0 = Q12 Q22 0 0
Q12 = 3.168 GPa, Q26 = 0,

0 0 , Q66 Q22 = Q12 0
Q16 = 0, Q66 = 4.5 GPa.

Q12 Q11 0 0 0 . Q66
The stiffness matrices of the layers are then given by:

Q90
90 0 90
1 mm 1 mm 1 mm h = 3 mm
FIGURE 25.1. Symmetric cross-ply laminate.
558
Relation (25.17) allows us to express the strains of thermal origin in the 0:

* 1 0 xx * L yy = 0 1 T . T * 0 0 xy 0
Thus:
* L T xx * yy = T T . * 0 xy 0
(25.24)
Similarly, for the 90 layers, we have:

* T T xx * yy = L T . * 0 xy 90
(25.25)
The resultants of thermal origin, deduced from Relations (25.21), are written as:
h * Nx = ( Q11 + 2Q12 ) L + ( 2Q22 + Q12 ) T 3 T , h N* y = ( Q22 + 2Q12 ) T + ( 2Q11 + Q12 ) L 3 T ,
* N xy = 0 (resulting from Q16 = Q26 = 0 and * xy = 0).
(25.26)
The moments are zero as a result of the symmetry of the laminate:

* * Mx = M* y = M xy = 0 .
The numerical application leads to:

* Nx = 733.7 T ,
N* y = 806.7 T .
The strains and curvatures are determined by substituting the resultants and moments into the constitutive equation (25.20), which, in the absence of external mechanical loads acting on the laminate, are written as:
* Nx A11 * N y A12 * N xy 0 M* = 0 x M* 0 y * 0 M xy
A12 A22 0 0 0 0
0 0 A66 0 0 0
0 0 0 D11 D12 0
0 0 0 D12 D22 0
0 0 xx 0 0 yy 0 0 xy . 0 x 0 y D66 xy
(25.27)
559
Whence:
* 0 Nx = A11 xx + A12 0 yy , 0 0 N* 12 xx + A22 yy , y = A 0 xy
= 0,
(25.28)
x = y = xy = 0.
From this we deduce the strains in the middle plane:
0 * Nx N* xx = A11 + A12 y,
0 yy
with
= A11 = A22 A22
* Nx N* = A12 + A22 y,
(25.29)
, ,
= A12
A12
A11
2 = A11 A22 A12 .
The stiffness constants of the laminate are:

A11 = ( Q11 + 2Q22 ) A12 = 3Q12 h = 66.418 106 Nm 1, 3
h = 9.504 106 Nm 1 , 3 h A22 = ( Q22 + 2Q11 ) = 102.18 106 Nm 1. 3
Thus:
= 15.259 109 m/N, A11 = 1.4193 109 m/N, A12 = 9.919 109 m/N. A22
(25.30)
That leads to:

0 xx = 10.05 106 T ,
6 0 yy = 6.96 10 T .
(25.31)
The stresses in the layers are next determined from Relation (25.18). For the 0 layer, we have: xx Q11 Q12 yy = Q12 Q22 0 0 xy 0
0 0 xx L T 0 T 0 , T yy Q66 0
560
or
0 Q11 ( xx L T ) + Q12 0 yy T T xx 0 0 yy = Q12 ( xx L T ) + Q22 yy T T xy 0 0
( (
) . )

(25.32)
Whence: 190.9 103 T xx 3 117.2 10 T = . yy 0 xy 0 For the 90 layers:

0 Q22 ( xx T T ) + Q12 0 yy L T xx 0 0 yy = Q12 ( xx T T ) + Q11 yy L T xy 90 0
(25.33)
( (
) . )

(25.34)
Thus: 95.5 103 T xx = 58.6 103 T yy 0 xy 90 . (25.35)
For the variation of temperature considered: T = 100 C , the values of the stresses are: xx 19.1 MPa yy = 11.7 MPa , 0 xy 0 xx 9.6 MPa yy = 5.9 MPa . 0 xy 90 (25.36)
The stress state of thermal origin is represented in Figure 25.2. It should be noted that the stress in the 0 layer reaches the value of 11.7 MPa in the direction transverse to the fibres, of the order of a quarter to a third of the fracture stress in this direction. It appears thus that the stresses of thermal origin, related to the mode of elaboration (curing at a temperature higher that the working temperature), must be taken into consideration in some design processes.
25.3.2.2 Thermal Expansion of a Symmetric Angle-Ply Laminate

In the case of a layer, referred to the directions (x, y) making an angle with the direction L of the materiall (Figure 25.3), the strains of thermal origin can be written, according to Relation (25.17), as:
561
y
T = 11.7 MPa
T = 5.9 MPa
x
L = 19.1 MPa
x
L = 9.5 MPa
0 layer
90 layer
FIGURE 25.2. Stresses of thermal origin in the layers of the symmetric cross-ply laminate of Figure 25.1.
2 * xx cos * 2 yy = sin * sin 2 xy
sin 2 L cos 2 T . T sin 2

1 * xy = xy T ,
(25.37)
The strains can thus be written in the form:

1 * xx = x T , 1 * yy = y T ,
(25.38)
on introducing the coefficients of thermal expansion referred to the material directions of the layer:
2 2 1 x = L cos + T sin , 2 2 1 y = L sin + T cos ,
(25.39)
1 xy = ( L T ) sin 2 .
y L x
y L x
single layer
laminate L
FIGURE 25.3. Single layer and angle-ply laminate.
562
In the case of the layer considered in the previous subsection, the coefficient of thermal expansion in the x direction is expressed:
2 2 6 /C . 1 x = ( 5cos + 20sin ) 10
(25.40)
The variation of 1 x as a function of the angle is reported in Figure 25.4. In the case of a symmetric angle-ply laminate made of n layers, the stiffness constants of the layers at orientation are:
= Q11 + , Q11 = Q22 + , Q22
* * xx = xx + ,
= Q12 + , Q12 = Q26 + , Q26

* * yy = yy + ,
= Q16 + , Q16 = Q66 + , Q66

* * xy = xy + .
(25.41)
and the strains of thermal origin are related by: (25.42)
The resultants and moments of thermal origin, deduced from Relations (25.21), are written as: * 1 1 1 = h Q11 Nx x + Q12 y + Q16 xy T ,
( ) 1 1 1 N* y = h ( Q12 x + Q22 y + Q26 xy ) T ,
(25.43)
* = 0, N xy
the stiffnesses where h is the thickness of the laminate and denoting by Qij
20 18 16 14 12 10 8 6 4 0 10 20 30 40 50 60 70 80 90
Thermal expansion x ( 106/C )
single layer laminate
Orientation ( )
FIGURE 25.4. Variation of the coefficient of thermal expansion of a single layer compared with that of a symmetric angle-ply laminate.
563
+ of the layer of direction . The symmetry of the laminate implies that the Qij
moments are zero:

* * Mx = M* y = M xy = 0.
The strains and curvatures are deduced from the constitutive equation (25.20):
* 0 Nx = A11 xx + A12 0 yy , 0 0 N* 12 xx + A22 yy , y = A 0 = 0, xy
(25.44)
x = y = xy = 0.
From this we deduce the strains in the middle plane:

0 xx = * A22 N x A12 N * ( y ),
1 1
0 yy =
with
( A12 N x* + A11N *y ) ,
(25.45)
2 = A11 A22 A12 .
The stiffness constants of the laminate are: , A11 = hQ11 , A12 = hQ12 . A22 = hQ22
Whence the expression of the elongation per unit length in the x direction:
0 xx = 1 x+
Q16 Q12 Q26 1 Q22 xy T . Q22 Q12 2 Q11
(25.46)
The coefficient of thermal expansion of the laminate in the x direction is thus expressed as: Q16 Q12 Q26 1 Q22 n (25.47) x = 1 xy . x+ Q22 Q12 2 Q11
n The variation of the coefficient of thermal expansion x of a symmetric angleply laminate as a function of the angle of the laminate is compared in Figure 25.4 with the coefficient of thermal expansion 1 x of a single layer.
25.3.3 Fundamental Relations

The fundamental relations of the mechanical behaviour of laminates, in the presence of expansional phenomena, are obtained by substituting the constitutive equation (25.20) into the fundamental relations (13.57) of the behaviour of plates without transverse shear, or into Equations (23.13) to (23.15), in order to take
564
account of buckling. The strains being expressed by Relations (14.15) as functions of the displacements, we finally obtain:
A11 2 u0 x 2 + 2 A16 2u0 2u0 2v0 2v0 2v0 + A66 + + + + A A A A ( ) 16 12 66 26 xy xy y 2 x 2 y 2 3w 0 x 2y ( B12 + 2 B66 ) 3w 0 xy 2 B26 3w 0 y 3
B11
3w 0 x3
3B16
* N * N x xy x y
= s
2 u0 t 2
(25.48)
2u0 2u0 2v0 2v0 2v0 2 + A26 + + + A A A 66 26 22 xy xy y 2 x 2 y 2
3w 0 x 2 y 3B26 3w 0 xy 2 B22 3w 0 y 3
A16
2u0 x 2
+ ( A12 + A66 )
B16
3w 0 x3
( B12 + 2 B66 )
* N xy
N * y y
x = s
2v0 t 2 4w 0 x 4 3u0 x3
,
4w 0 x3y 3u0 x 2y + 2 ( D12 + 2 D66 ) ( B12 + 2 B66 ) 3v0 xy 2 4w 0 x 2y 2 4w 0 xy 3 B16
(25.49)
4w 0 y 4
D11
+ 4 D16 3B16
+ 4 D26 3u0 y 3
+ D22 3v0 x3
B11
3u0 xy 2 y 3
B26
( B12 + 2 B66 )
3v0 x 2y
3B26
B22
3v0
* 2M x
x 2
+2
* 2 M xy
xy
2M * y y 2
+ s
2w 0 t 2 (25.50)
i = q + Nx
2w 0 x 2
i + 2 N xy
2w 0 2w 0 . + Ni y xy y 2
These relations differ from the fundamental relations (16.1), (16.2) and (23.21) by the presence of the resultants and moments induced by the expansion phenomena. * , * ) are independent of x and y (as In the case where the expansional strains (11 22 the type of problem considered in Section 25.4), the resultants and moments, induced by the expansion processes, are also independent of the variables x and y,
565
and do not then appear explicitly in the fundamental relations (25.48) to (25.50). However, the boundary conditions imposed induce in-plane loads, which act as i i pre-buckling resultants N x , N iy , N xy . It results that these in-plane loads modify the structure behaviour: flexural behaviour, vibrations and buckling.
25.3.4 Strain Energy

In the case of a variational formulation of the fundamental relations of laminates, the expressions obtained in Section 16.3 show that the expansion phenolmena influence only the expression of the strain energy. In the presence of the expansion phenomena, Expression (16.33) is modified as follows:
Ud = 1 2
xx
* * ( xx * xx ) + yy ( yy yy ) + zz ( zz zz )
+ yz yz * yz
+ xz xz * xz
+ xy xy * xy
) d x d y d z.
(25.51)
Taking account of the assumptions of the classical laminate theory: zz = 0, * = * = 0, and of Expression (25.18) for the stresses in each layer, xz = yz = xz yz Expression (25.51) can be written in the form:
Ud = 1 2
k k ) + Q k k Q11 ( xx xx 22 yy yy 2
k k + Q66 xy xy
( ) k k + 2Q26 ( yy yyk )( xy xy ) d x d y d z,
k k ) k + 2Q k ( k ) k + 2Q12 ( xx xx 16 xx yy yy xx xy xy
(25.52)
k ) of layer k are denoted Qij where the off-axis stiffness constants ( Qij . In the prek
sence of expansion processes, Expression (25.52) of the strain energy replaces Expression (16.34). By introducing the strain-displacement relations (14.14) and (14.15) into Expression (25.52), and then integrating with respect to z through the thickness of the laminate, we obtain:
U d = U d ( = 0) + +
v0 u0 v 0 u0 N x x + N y y + N xy y + x d x d y
2 2 2w 0 w0 w0 2 M + M + M x dx dy y xy xy x 2 y 2 h/2 f ( ik ) d x d y, h/2
(25.53)
where U d ( = 0) is the strain energy, in the absence of the expansion processes,
566
expressed by (16.35), and:

k k ) + Q k k ( xx f ( ik ) = Q11 yy 22 2
( )
k k + Q66 xy
( )
k k k xx yy 2Q12
k k k xx xy + 2Q16
k k k yy xy . + 2Q26
(25.54)
As the function f ( ik ) is independent of the displacements u0 , v0 and w 0 , the integral involving this function will vanish in the expression for the first variation Ud of the strain energy.
25.4 BEHAVIOUR OF RECTANGULAR PLATES 25.4.1 Rectangular Plate Made of a Symmetric Laminate
In this Section, we analyse the influence of the expansion phenomena upon the flexural behaviour, the vibrations and the buckling of a rectangular plate made of a symmetric laminate. This type of laminate is characterized by:
Bij = 0, i, j = 1, 2, 6.
(25.55)
The presence of the coefficients D16 and D26 of bending-twisting coupling does not allow in this case to solve the fundamental relations (25.48) to (25.50) directly, and approximate solutions can be obtained using the Ritz method. We consider the case of initial in-plane resultants induced by expansion processes, resulting from the conditions imposed at the boundaries. If the expansional strains are independent of coordinates (x, y) and are even functions of the coordinate z, Relations (25.21) show that the resultants owed to the expansional effects are constant and the moments are zero: M x = M y = M xy = 0 . For symmetric laminates (Section 22.1), the displacements u0 and v0 are zero. From this it results that the strain energy (25.53) is identical to the strain energy expressed by (22.2): 1 Ud = 2
2 2 2w 0 2w 0 2w 0 2w 0 D11 + 2 D12 + D22 2 2 2 2 x x y x =0 y = 0 y (25.56) 2 2 2 2 2 w0 w0 w0 w0 + 4 D16 + D26 + 4 D66 d x d y. 2 2 xy x y xy a b
The energy function:

U = U d Wm + Wf Ec
(25.57)
is deduced from Equations (25.56), (16.42), (16.45) and (23.33), which leads to
25.4 Behaviour of Rectangular Plates
567
the result:
1 U= 2

x =0
b y =0
2 2 2w 0 2w 0 2w 0 2w 0 D11 + D22 + 2 D12 2 x 2 x 2 y 2 y 2 2
2w 0 2w 0 2w 0 2w 0 2w 0 + 4 D66 + D26 + Nx 2 + 4 D16 x 2 y 2 xy xy x 2 2 2w 0 w 0 w 0 2 ( ) 2 w q + N y 2 + 2 N xy + 0 d x d y, s x y y (25.58)
where Nx, Ny and Nxy are the in-plane resultants (as defined in (23.23)) induced by the expansion phenomena in conjunction with the conditions imposed at the boundaries. The approximate solution is sought in the usual form of a double series:
w 0 ( x, y ) =
Amn X m ( x)Yn ( y) ,
m =1 n =1
(25.59)
Where the functions Xm(x) and Yn(y) have to satisfy the conditions imposed along the edges x = 0, x = a and y = 0, y = b . The coefficients Amn are determined by the stationnarity conditions (8.66):
U = 0, Amn
m = 1, 2, . . . , M , n = 1, 2, . . . , N ,
(25.60)
is the energy obtained by substituting the approximate expression where U (25.59) for the deflection into Expression (25.58). Conditions (25.60) then lead to the M N equations:
2200 2002 0220 2 0022 4 + D12 ( Cminj + Cminj ) + 4 D66C1111 minj {D11Cminj R + D22Cminj R
i =1 j =1
1210 2101 0121 3 + 2 D16 Cminj + Cminj R + 2 D26 C1012 minj + Cminj R 1100 0011 2 1001 0110 + a2 N xCminj + N y Cminj R + N xy Cminj + Cminj R 0000 0 0 Aij = a 4 q0 I m Jn , s a 4 2Cminj
pour m = 1, 2, . . . , M ,
n = 1, 2, . . . , N ,
(25.61)
where the coefficients
pqrs Cminj
were introduced in (21.116).
The system of equations so obtained regroups and generalizes the equations for bending of plates (Equations (21.119), (22.5)), the buckling equations (Equation (23.142)), and the equations for the flexural vibrations (Equations (24.109) and (24.146)). The system of equations applies as well to the case where the plate is
568
subjected to expansional strains independent of (x, y) as in the case where there are no expansion processes. In this latter case, the resultants Nx, Ny and Nxy are the initial in-plane resultants imposed at the boundaries (other than by expansion processes), the imposed moments being zero. When the expansion phenomena are taken into account, the resultants Nx, Ny and Nxy are the in-plane resultants induced by the expansional strains and by the boundary constraints imposed along the edges of the plate. As a function of the various analyses considered in the preceding chapters, the beam functions used for the study of the vibrations of beams (Section 24.3) can be chosen as functions Xm(x) and Yn(y) for expressing the approximate solutions (25.59). In the case of a plate subjected to a static transverse load ( N x = N y = N xy = 0, = 0) , solving the system (25.60) leads to the determination of the coefficients Aij. In the absence of transverse loads (q = 0), Equations (25.61) constitute a system of homogeneous equations. A nonzero solution ( Aij 0) is then obtained when the determinant of the matrix of the coefficients Aij vanishes. This condition allows us to determine the natural frequencies of the vibrations of the plate subjected or not to initial in-plane loads, imposed or not by expansion processes. This condition also allows us to determine the critical buckling load (resulting or not from expansion processes), which corresponds, in the case where = 0, to the combination of the lowest value of the resultants Nx, Ny and Nxy which makes the determinant vanish.
25.4.2 Rectangular Plate Made of an Antisymmetric Angle-Ply Laminate

An antisymmetric angle-ply [ ]p laminate, comprising an even number of layers, is characterized (Relation (22.54)) by:
A16 = A26 = 0, B11 = B12 = B22 = B66 = 0, D16 = D26 = 0 .
(25.62)
In the case where the expansional strains are independent of coordinates (x, y) and are even functions of z in each layer, Relations (25.21) and (25.17) show that Nx , N y et M xy are constant, whereas N xy = M x = M y = 0 . In the case of bending in presence of initial in-plane loads, the fundamental relations (25.48) to (25.50) then reduce to:
A11 2 u0 x 2 + A66 2 u0 y 2 + ( A12 + A66 ) 2v0 3w 3w 0 3B16 2 0 B26 =0, xy x y y 3
(25.63)
2u0 2v0 2v0 3w 0 3w 0 + A66 2 + A22 2 B16 3B26 = 0, ( A12 + A66 ) xy x y x3 xy 2
(25.64)
569
D11
4w 0 x 4
+ 2 ( D12 + 2 D66 ) 3B26 3v0 xy 2
4w 0 x 2y 2
+ D22
4w 0 y 4
3B16
3u0 x 2y
B26
3u0 y 3
B16
3v0 x3
i = q + Nx
2w 0 x 2
i + 2 N xy
2w 0 2w 0 . + Ni y xy y 2
(25.65)
From the constitutive equation (25.20), in conjunction with Expressions (14.15), which give the strains as functions of the displacements, the in-plane resultants are expressed as:
Nx = Nx + A11
u0 v 2w 0 + A12 0 2 B16 , x y xy u0 v 2w 0 + A22 0 2 B26 , x y xy
(25.66) (25.67) (25.68)
N y = N y+A 12
N xy
2w 0 2w 0 u0 v0 = A66 + B16 B26 . y x 2 y 2 x
The resultants N x and N y , owed to the expansion processes, are induced by
the expansional strains and the constraints imposed at the boundaries, whereas the other terms result from the strains induced by the bending load q. In the case where we are interested only in the effect of in-plane loads resulting from the expansion phenomena, the initial in-plane loads, appearing in Equation (25.65), are given by:
i , Nx = Nx
Ni y = N y ,
i N xy = 0.
(25.69)
We examine the case of hinged edges free in the edge directions. The boundary conditions are then expressed according to Relations (22.56) to (22.62). The transverse load q = q(x, y) is expanded as a double Fourier series:
q ( x, y ) =
with
qmn =
qmn sin m a sin n b ,

m =1 n =1
b y =0
(25.70)
4 ab

x =0
q ( x, y ) sin m
(25.71)
The solutions of the problem are then sought by expanding the displacements as double Fourier series satisfying the boundary conditions (22.59) to (22.62):
570

u0 ( x, y ) =
Amn sin m a cos n b ,

m =1 n =1

(25.72)
v 0 ( x, y ) =
Bmn cos m a sin n b ,

m =1 n =1

(25.73)
w 0 ( x, y ) =

m =1 n =1
(25.74)
On substituting Expressions (25.72) to (25.74) into Equations (25.63) to (25.65), and then solving the system of equations so obtained, we find:
Amn = qmn a3
3 2 2 2 2 2 2 nR ( m A66 + n R A22 )( 3m B16 + n R B26 )
mn
a3
m 2 ( A12 + A66 ) ( m 2 B16 + 3n 2 R 2 B26 ) ,

2 2 2 2 2 2 m ( m A11 + n R A66 )( m B16 + 3n R B26 )
Bmn = qmn
mn
a4
(25.75)
n 2 R 2 ( 3m 2 B16 + n 2 R 2 B26 ) ( A12 + A66 ) , ( m 2 A11 + n 2 R 2 A66 )( m 2 A66 + n 2 R 2 A22 )
Cmn = qmn
mn
m 2 n 2 R 2 ( A12 + A66 )2 ,
with
mn
a2 2 4 2 2 2 4 4 = m D11 + 2m n R ( D12 + 2 D66 ) + n R D22 2 ( m N x + n 2 R 2 N y )

2 2 2 2 2 2 2 2 2 2 ( m A11 + n R A66 )( m A66 + n R A22 ) m n R ( A12 + A66 )
+ 2m 2 n 2 R 2 ( A12 + A66 ) ( 3m 2 B16 + n 2 R 2 B26 )( m 2 B16 + 3n 2 R 2 B26 ) m 2 ( m 2 B16 + 3n 2 R 2 B26 )
( m2 A11 + n2 R 2 A66 ) 2 n 2 R 2 ( 3m 2 B16 + n 2 R 2 B26 ) ( m 2 A66 + n 2 R 2 A22 ) .

2
(25.76)
Expressions (25.75) have a form identical to Expressions (22.67) obtained in the case of bending under the single load q. They differ only by the introduction into the expression for mn of the resultants N x and N y induced by the expansion phenomena. Moreover, the critical buckling load corresponds to the combination of the smallest value of the resultants N x and N y , making mn zero, the coefficients Amn, Bmn and Cmn then being undetermined.
571
25.4.3 Thermal Effects

In the preceding Subsections 25.4.1 and 25.4.2, we have taken into account the expansion phenomena without considering their nature (thermal, swelling, etc.). To illustrate the results obtained, we examine in this subsection the effects induced by the thermal expansion processes, in, the case of a plate made of a symmetric angle-ply laminate of type [()p]s. This laminate is characterized (Relations (15.25)) by:
A16 = A26 = 0,
Bij = 0,
D16 = D26 = 0 .
(25.77)
In the case where the plate is clamped along the edges x = 0 and x = a, and free along the edges y = 0 and y = b, the boundary conditions are: along the edges x = 0 and x = a :
w 0 = 0, along the edges y = 0 and y = b :

M y = 0, 2
w 0 = 0, x
M xy x M y y
(25.78)
= 0.
(25.79)
The conditions (25.79) along the free edges are deduced from conditions (16.32) and Relations (13.56) of plates. These conditions lead, for y = 0 and y = b, to:
D12 3w 0 x3 + ( D12 + 4 D66 ) 2w 0 x 2 2 D26 + 4 D26 2w 0 2w 0 D22 = 0, xy y 2 3w 0 xy 2 + D22 3w 0 y 3 =0.
(25.80)
2 D16
3w 0 x 2y
(25.81)
In agreement with the results established in Subsection 25.4.1, the bending of the plate is studied by the Ritz method, by introducing the beam functions satisfying the imposed boundary conditions: for the clamped edges x = 0 and x = a : x x x x X m ( x) = cos m cosh m m sin m sinh m , a a a a for the free edges y = 0 and y = b : Y1 ( y ) = 1, y 3 1 2 , b y y y y Yn ( x) = cos n + cosh n + n sin n + sinh n , n 3. b b b b Y2 ( y ) = (25.83)
(25.82)
572
The values of the coefficients i et i are reported in Table 21.3 for the clampedclamped beam function and in Table 24.3 for the free-free beam function. The values of the integrals appearing in Expression (25.60) are reported in the tables of Appendix B for the clamped-clamped beam function. The values of the integrals are to be evaluated in the case of the free-free beam function. These values allow (Subsection 25.4.1) either to describe the plate bending in determining the coefficients Aij by solving the system (25.61), or to determine the critical buckling load by making the determinant of the system (25.61) vanish. When determining the critical buckling load, the in-plane loads Nx, Ny and Nxy appearing in Equation (25.61) can be expressed by the constitutive equation (25.20), that is: N x A11 N y = A12 N xy 0 A12 A22 0
0 * 0 xx Nx 0 * 0 yy N y . 0 * A66 xy N xy
(25.84)
Furthermore, the boundary conditions impose: along the clamped edges x = 0 and x = a :
0 xx = 0,
(25.85)
along the free edges y = 0 and y = b : N y = N xy = 0 . Combining the preceding results leads to:
+ Nx = Nx
(25.86)
A12 Ny , A22
(25.87)
or by considering Expressions (15.25):

+ Nx = Nx
Q12 N y. Q22
(25.88)
The resultants induced by the thermal effects are written (Relation (25.21)) as:
* Nx
* * * xx + Q12 yy + Q16 xy )k d z , h (Q11

k =1 n
k 1
hk
(25.89)
N* y =
* * * xx + Q22 yy + Q26 xy )k d z , h (Q12

k =1
k 1
hk
(25.90)
where, taking into account Relations (25.1) and (25.17), the thermal strains referred to the reference directions of the plate are given by:
Exercises
573
* cos 2 xx * 2 yy = sin * 2sin cos xy
L cos 2 T , T 2sin cos sin 2
(25.91)
on introducing the coefficients of thermal expansion of a layer, referred to the material directions (L, T) of the layer. Thus:
2 2 * xx = ( L cos + T sin ) T , 2 2 * yy = ( L sin + T cos ) T ,
(25.92)
* xy = ( L T ) sin 2 .
By substituting these expressions into Relations (25.88) to (25.90), and then taking into account the symmetries of the laminate, we obtain: 2 Q12 2 2 + N x = Q11 ( L cos + T sin ) Q11 Q + Q26 12 ( L T ) sin 2 hT , + Q16 Q11
(25.93)
where h is the thickness of the laminate. Substitution of the expression for Nx into the system of equations (25.61) next allows us either to study the influence of the temperature upon the natural frequencies of the vibrations of the plate, or to determine the critical temperature which induces the buckling of the plate. A similar approach can also be applied for analysing the effects of the hygrothermic phenomena.
EXERCISES
24.1 A symmetric laminate material is made of three layers. Layers 1 and 3 are unidirectional reinforcement layers of thickness 1.2mm, with the mechanical characteristics:
EL = 46 GPa,
ET = 10 GPa,
LT = 0.30,
GLT = 4.8 GPa,
and coefficients of thermal expansion:
L = 5 106 /C,
T = 22 106 /C.
Layer 2 is a double mat reinforcement layer of thickness 2.8mm, with the mechanical characteristics: EL = ET = 8 GPa,
LT = 0.32,
GLT = 3.2 GPa,
and coefficients of thermal expansion:
574
L = T = 18 106 /C
Curing of the laminate is carried out at a temperature of 125C. We study its mechanical state at the working temperature of 22C in the reference directions (x, y), which are the same as the material directions (L, T) of layers 1 and 3. Derive the strains in the layers and the in-plane resultants induced by the thermal processes. From these results deduce the thermal in-plane strains of the laminate. Then derive the thermal in-plane stresses in each layer.
25.2 A nonsymmetric laminate material is made of two layers. Layer 1 has the same characteristics as layers 1 and 3 of Exercise 25.1. Layer 2 has the same mechanical and thermal characteristics as layer of Exercise 25.1, but is half the thickness (i.e. 1.4 mm). Curing is carried out at the same temperature of 125C and we consider its mechanical state at 22C. Derive the thermal strains in the layers, the in-plane resultants and moments, induced by the thermal processes. From these results deduce the in-plane strains and the curvatures of the laminate. Derive the deflection observed, after demoulding, on a plate of length a et and width b. Find the in-plane stresses in each layer. 25.3 A plate made of the laminate material considered in Exercise 25.1 is clamped, at a temperature of 22C, along its four edges, parallel to the material directions. The plate is then raised to a temperature of 50C. Derive the strains in the layers and the in-plane resultants, at the temperature of 50C. Then deduce the in-plane strains. Determine the in-plane stresses in each layer. 25.4 A beam, of length L, made of the laminate material considered in Exercise 25.1 is clamped at its ends. We study the buckling and the bending vibrations of the beam, at a temperature of 22C. Find the critical buckling load of the beam. Derive the natural frequencies of the bending vibrations of the beam. In the two cases, compare the results obtained with the case in which no thermal expansion would be induced.
CHAPTER 26
Predesigning Laminate and Sandwich Structures
26.1 PROBLEM OF DESIGNING

Parts 1 and 3 of this book clearly show how the engineer is now able to tailor composite materials, in such a way to obtain the desired properties by making an appropriate choice of the constituents (fibres and matrix), the proportion of the reinforcement, the form of the reinforcement (unidirectional, cloth, mat, etc.), the nature of the fibre-matrix interface (high or poor adhesion), the stacking sequence of layers, the type of composite (laminate or sandwich), etc. Thus, the tailorability of composite materials considerably modifies the conventional approach to the design of a composite material structure. The concepts of material and structure, which are distinct and independent in the traditional design processes, become intimately linked in the designing of composite material structures [32, 33]. To design a structure which is to be elaborated from a composite material involves designing the most suitable material at the level of each layer as well as the level of laminate or sandwich material, in close relation with the process of the structure design. So, in the process of design and optimisation of composite material structures, the design of the material plays a fundamental role and consists in: (i) the design of every layer by optimising their properties according to the constituents (the nature of the fibres and matrix), the proportion of fibres, the fibre-matrix interface, the arrangement of the fibres, etc., (ii) the design of the laminate or sandwich material by optimising its properties according to the properties of the layers and by the arrangement of the stacking sequence of the layers. Owing to these peculiar characteristics of composite materials, no fixed entry point in the design process can be strictly defined. The performances of the structure depend of the properties of the composite material which is used, properties which in turn can be properly tailored. So, a circular process for designing is involved from constituents to materials, from materials to structure, and so on. Finally, the design is not limited to structure considerations, but
576
Chapter 26 Predesigning Laminate and Sandwich Structures
includes the tailoring of the materials as well as the manufacturing technologies. Thus, the property of tailoring the composite materials enriches at infinity the possibilities of designing the composite material structures, however at the cost of a great complexity of the design processes which can be efficiently implemented only with advanced numerical tools. The problem of predesigning composite material structures can be solved by analytical approaches or numerical methods. The advantage of analytical evaluations lies in their general application, which allows to take into account easily the influence of the various design. The applications of the analytical approaches (considered in Part 4) are, however, restricted to the analysis of simple elements of composite material structures (Part 5), such as beams and plates. Moreover, they constitute the necessary and indispensable introduction to the problem of the design of complex shape structures by using numerical methods, these methods being based on the analytical models considered in Part 4. In fact, we have noted that the ability of tailoring the composite materials introduces a considerable complication in the design of materials and structure, requiring a high level of knowledge of the analysts and designers. So, an efficient predesigning needs advanced tools for the analysis and requires to resort to computer in the very early stages of the conception. Schematically, the mechanical properties of the layer can be obtained by experimental tests or derived from analytical procedures as functions of the properties of the constituents (Part 3). The mechanical behaviour of a laminate material or sandwich material is evaluated by the different models (Part 4), from which the numerical analyses are implemented. Next, the optimisation of the design of complex structures complexes made of composite materials combines computer aided design with finite element analysis.
26.2 BASIC ELEMENTS OF COMPOSITE STRUCTURES 26.2.1 Simple Beams

26.2.1.1 Layering Plane Orthogonal to the Loading
In the case where the loading is orthogonal to the plane of layering (Figure 26.1), the differential equation of a beam made of a symmetric laminate is (20.10) expressed as: d 2w 0 dx
2
M , Ex I
(26.1)
26.2 Basic Elements of Composite Structures
577
introducing the quadratic moment of the cross-section of the beam:
I = I xy = and the bending modulus of the beam:

Ex =
bh3 , 12
(26.2)
12
h3 D11
(26.3)
with
= D11
2 ( D22 D66 D26 ),
2 2 2 D22 D16 D66 D12 = D11D22 D66 + 2 D12 D16 D26 D11D26 .
The stiffness coefficients are expressed by Relations (14.27) or (14.33): 1 3 3 ) = hk Dij = hk 1 ( Qij k 3 k =1
k . (Qij )k ek zk2 + 12 k =1
e3
(26.4)
The beam theory makes the assumption (20.7) that the deflection w0 is a function only of x. This assumption is satisfied in the case where the length-to-width ratio (L/b) of the beam is sufficiently high. Equation (26.1) is analogous to the equation of the classical theory of isotropic beams and may be written in the form: d 2w 0 dx
2
M , Jx
b
D11
(26.5)
where Jx is the bending stiffness of the beam in the x direction, expressed as:
J x = Ex I =
(26.6)
h x b
FIGURE 26.1. Beam loaded orthogonally to the plane of lamination.
578
In the case of orthotropic laminates (D16 = D26 = 0), the bending modulus and the bending stiffness are expressed as:
Ex =
2 12 D12 D 11 , D22 h3
(26.7)
and
2 D12 J x = b D11 . D22
(26.8)
2 /D22 can be neglected compared with D11, the In the case where the term D12 modulus and the stiffness reduce to: D , (26.9) E x = 12 11 h3 and
b 3 3 )k . hk J x = bD11 = hk 1 ( Q11 3 k =1
(26.10)
26.2.1.2 Lamination Plane in the Loading Plane

In the case of loading in the plane of lamination (Figure 26.2), the plane stress state in layer k may be written (11.43) as:
Q12 xx Q11 Q22 zz = Q12 Q xz k Q16 26 xx Q16 zz . Q26 Q66 xz
(26.11)
On expressing the strain field in the form introduced in (14.14):

0 xx xx x 0 zz = zz + z z , 0 xz xz xz
(26.12)
and introducing the bending and twisting moments per unit length of beam defined, in analogy with (13.19), as follows:
Mx 1 Mz = h M xz
y =h
k =1
hk
k 1
xx z zz d y d z , z =b 2 xz k
b2
(26.13)
the bending constitutive equation along the x direction, corresponding to zero in0 0 0 = zz = xz = 0) , can be written, in the case of a symmetric lamiplane strains ( xx nate, on combining Equations (26.11) to (26.13), in the form:
579
b x h
FIGURE 26.2. Beam loaded in the plane of lamination.
Mx A11 1 b3 M z = 0 = h 12 A12 M = 0 A xz 16
n
A12 A22 A26

n
A16 x A26 z , A66 xz
(26.14)
where the coefficients Aij are the in-plane stiffness constants introduced in (14.23): Aij =
( hk hk 1 ) (Qij )k = ( Qij )k ek .
k =1 k =1
(26.15)
In the case of bending in the plane of lamination, Equation (26.14) thus replaces Equation (20.1) derived for bending orthogonal to the plane of lamination. By analogy with the results obtained in Subsection 20.2.1, the bending differential equation is written in the form (20.10) or (26.1): d 2w 0 dx
2
M , Ex I
hb3 , 12
(26.16)
where the quadratic moment of the cross-section of the beam is given by:
I = I xz = and the bending modulus is expressed as: (26.17)
580
Ex =
1
hA11
(26.18)
with
A11 =
2 ( A22 A66 A26 ),
2 2 2 . A22 A16 A66 A12 = A11 A22 A66 + 2 A12 A16 A26 A11 A26
Equation (26.16) can also be written in the form (26.5): d 2w 0 dx

2
M , Jx
b3
(26.19)
by introducing the bending stiffness of the beam expressed as: . (26.20) 12 A11 In the case of orthotropic laminates (A16 = A26 = 0), the bending modulus and the bending stiffness are given by:
2 1 A12 E x = A11 , h A22
J x = Ex I =
(26.21)
and
2 b3 A12 J x = A11 . A22 12
(26.22)
2 In the case where the term A12 /A22 can be neglected compared to A11, the modulus and the stiffness reduce to: A (26.23) E x = 11 , h and
b3 b3 Jx = A11 = 12 12
)k ek . (Q11
k =1
(26.24)
26.2.2 Profiles
Most standard profiles are constructed of flanges and webs of orthotropic thin plates, in which the laminate is both balanced and symmetrical. The bending differential equation can again be written in the form (26.5), (26.19), as: d 2w 0 dx
2
M , Jx
(26.25)
where the bending stiffness Jx combines the results obtained in Subsections 26.2.1.1 and 26.2.1.2. We consider two examples hereafter.
581
1. I Profile We consider a beam with a constant I-shaped cross-section reported in Figure 26.3, constructed with laminate of mat layers and unidirectional layers of 1 mm thickness, and having the mechanical characteristics:
unidirectional layers (UD):

LT = 0.32,
GLT = 3.6 GPa,
mat layers (M):

EL = ET = 7.5 GPa,
LT = 0.33.
The stiffness constants of the layers are: unidirectional layers:

UD Q11 = 38.945 GPa, UD Q22 = 9.224 GPa, UD Q12 = 2.952 GPa, UD Q66 = 3.6 GPa, M Q12 = 2.777 GPa, M Q66 = 2.820 GPa.
mat layer:
M Q11 = 8.417 GPa, M Q22 = 8.417 GPa,
The in-plane stiffness coefficients of the laminate are given by:

UD M Aij = ( 4Qij + 3Qij ) 103 .
So: A11 = 181.031106 N/m, A22 = 62.147 10 N/m,

6
A12 = 20.139 106 N/m, A26 = 0,
A16 = 0, A66 = 22.860 106 N/m.
b = 100 mm
UD UD M M M UD UD
7 6 5 4 3 2 1
ht = 100 mm
UD UD M M M UD UD
50 mm 49 48 47 46 45 44 43
FIGURE 26.3. Beam with an I Profile.
582
The bending stiffness coefficients of a flange, referred to the middle plane are:
Dij = 1 3 3 3 3 UD 3 3 M { (45 43 ) + (50 48 ) Qij + (45 43 )Qij } 3 1 UD M = ( 26,026Qij + 19,467Qij 109. ) 3
So: D11 = 392.479 103 Nm, D22 = 134.639 10 Nm,

3
D12 = 43.630 103 Nm, D26 = 0,
D16 = 0, D66 = 49.530 103 Nm.
The bending stiffnesses of the flanges 1 and 2, evaluated by Expression (26.8), are expressed as: 2 2 J1 (26.26) x = J x = 37,834 Nm . The bending stiffness of the web, determined by Expression (26.22), is:
3 Jx = 9,250 Nm 2 .
The bending stiffness of the I profile is then:

2 3 2 J x = J1 x + J x + J x = 84,973 Nm .
(26.27)
Using the approximate relations (26.9) and (26.24), the estimated stiffness is:
J x = 88,091 Nm 2 ,
(26.28)
that is an error less than 4 %.

2. Square Profile
A calculation similar to the preceding one can be carried out in the case of the square profile of Figure 26.4. The difference from the I profile lies in the doubling of web walls in the plane loading. With the same constitution for the laminate as previously, the bending stiffness is:
2 3 2 J x = J1 x + J x + 2 J x = 94,223 Nm .
(26.29)
100 mm
100 mm
FIGURE 26.4. Square profile.
583
26.2.3 Sandwich Beams

The bending of sandwich beams has been studied in Section 20.4. Bending can be analysed (Subsection 20.4.2) either by the sandwich theory or by laminate theory including the transverse shear effect. In both cases, the bending differential equations (20.86) to (20.90) introduce the two functions w0 and x. In the case where the length-to-height ratio is high enough, the bending of sandwich beams can be approximated by the classical laminate theory. The bending differential equation may then be written in the classical form (26.5), the bending stiffness being expressed by (26.6) or (26.8). As an illustration, we consider the sandwich beam of Figure 26.5 the skins of which are made of the laminate considered in Subsection 26.2.2 (Figure 26.3) and with a foam core with the mechanical characteristics:
Ec = 200 MPa,
c = 0.40.
The stiffness matrix of the core is (20.147): 0 238.095 95.238 c Qij 0 MPa . = 95.238 238.095 0 0 71.429 According to the results obtained in Subsection 26.2.2, the stiffness coefficients are given by: 2 UD M c Dij = 26,026Qij + 19,467Qij + 79,507Qij 109 . 3 Whence:
D11 = 797.578 103 Nm,

3
D12 = 92.307 103 Nm,
D16 = 0, D66 = 102.846 103 Nm.
D22 = 281.899 10 Nm, D26 = 0,
100 mm
foam
100 mm
FIGURE 26.5. Sandwich beam.
584
Whence the bending stiffness determined by Expression (26.8):

J x = 76,735 Nm 2 .
(26.30) (26.31)
Neglecting the foam stiffness, the bending stiffness of the beam is:
2 J x = 2J1 x = 75,668 Nm ,
where J 1 x is the stiffness of the skins obtained in Subsection 26.2.2. The foam thus does not participate much in the total stiffness of the beam.
26.2.4 Plates
The analysis of the linear behaviour of plates has been considered in Part 4. The study of the bending of plates constituted of laminates can be carried out by the classical laminate theory (Chapter 16) in the case of low thicknesses of the plates, or by the theory of laminates taking transverse shear effect into account (Chapter 17) in the case of thick plates. The analysis of sandwich plates can be implemented using the theory of sandwich plates (Chapter 18) in the case where the thickness of the skins are small, or (Subsection (20.4.2) by the theory of laminates taking the transverse shear effect into consideration in the case of thick skins. Buckling of beams and plates has been analysed in Chapter 23.
26.3 DETERMINATION OF THE CHARATERISTICS OF THE MECHANICAL BEHAVIOUR 26.3.1 Engineering Constants
The classical laminate theory requires the knowledge of four moduli per layer: EL, ET, LT and GLT. Taking the transverse shear effect into account requires in addition the knowledge of the transverse shear moduli: GLT (the same as GLT in the case of a unidirectional layer) and GTT . These moduli can be derived analytically from the mechanical characteristics of the constituents in the case of unidirectional layers (Chapter 9) and in the case of cloth or mat reinforcement layers (Chapter 15). Experimentally, the moduli EL, ET, LT and GLT can be measured in tensile tests (Section 11.4). The values of the transverse moduli GLT and GTT can be obtained from three-point bending tests (Subsection 20.3.2), performed in the L or T direction and for various span lengths between supports.
26.3.2 Fracture Characteristics

The fracture criteria require the knowledge (Chapter 12) of the strengths of each layer: Xt, Xc, Yt, Yc, S. These strengths are measured experimentally (Subsection 12.2.2) in tensile, compressive and shear tests. The values obtained,
26.3 Determination of the Characteristics of the Mehanical Behaviour
585
associated with a given fracture criterion, will then allows us to estimate the limit loading state that will be able to support the structure under consideration without damage. When a laminate is subjected to given loads, the strain and stress state can be derived in each layer from Relations (14.44), (14.14), (14.46), (14.19) and (14.48). The strains and stresses obtained can then be compared with the fracture criterion considered (Section 12.2). This comparison allows us to estimate the loading at which the first ply failure will occur, corresponding generally with the fracture of layers the fibre direction of which is orthogonal o the direction of loading. However, in the case of layers with different orientations, the laminate will be able to support an increasing loading as the number of layer fractures increases, although with decreasing stiffness up to the final fracture of the laminate. To illustrate this behaviour, we consider the case of a symmetric laminate (Figure 26.6), made of layers of the same thickness but having different orientations: [0/30/60/90/120/150]s. The characteristics of the layers are:
LT = 0.31,
GLT = 4.5 GPa.
The laminate is subjected to a strain state applied in the x direction. As fracture criteria, we consider three possible modes of fracture: tensile fracture in the L direction:
L X t = 1 400 MPa,
tensile fracture in the T direction:
( L > 0) , ( T > 0) ,
(26.32) (26.33) (26.34)
T Yt = 40 MPa,
shear fracture:
LT S = 70 MPa .
150 120 90 60 30 0 0 30 60 90 120 150
FIGURE 26.6. Symmetric laminate with layers of different orientations.
586
As the laminate is symmetric and balanced, the constitutive equation (14.29) reduces to: N x A11 0 = A12 0 0 or
0 N x = A11 xx + A12 0 yy , 0 0 = A12 xx + A22 0 yy , 0 = 0. xy
0 These equations allow us to obtain Nx and 0 yy as functions of xx . Then, the 0 is determined in each layer by Relations (14.20) stress state for every value of xx and (14.48). Hence:
A12 A22 0
0 0 xx 0 0 yy , 0 A66 xy
(26.35)
(26.36)
L T = T Qk LT k
0 xx 0 yy . 0
(26.37)
0 The process considered for establishing the curve Nx as a function of xx in a controlled displacement test is the following: 0 is expressed by Relation (26.36). Nx as a function of xx
The fracture state in each layer is determined in accordance with the criteria (26.32) to (26.34). After tensile fracture in the T direction of a layer (26.33) or shear fracture (26.34), the moduli ET and GLT of the layer are set equal to zero. After tensile fracture in the L direction of a layer (26.32), the moduli EL and GLT are set equal to zero. The curve obtained by using this process is reported in Figure 26.7: the first fracture is produced by transverse fracture of the layers oriented at 90, the second fracture by transverse fracture of the layers oriented at 60 and 120, the third fracture by shear fracture of the layers oriented at 30 and 150. The final fracture occurs by fracture in the L direction of the 0 layers, implying fracture in the L directions of layers oriented at 30 and 150. We observe that the first fracture corresponds to a low load (about 20 % of the final load), whereas before final fracture the stiffness is some 69 % of the initial stiffness. This example shows the progressive nature of the laminate damaging, which will be needed to be taken into account at the design stage of a laminate or sandwich structure.
26.4 Analysis of Structures using the Finite Element Method
587
500
Longitudinal resultant Nx / h ( MPa )
3rd failure
final fracture
476 MPa
400
1st failure 2nd failure
300
15.3 GPa 238 MPa
200
122 MPa 17 GPa
100
95 MPa
20.6 GPa 22 GPa
0.4 0.6
1.4
2
0 xx
3 3.1
(%)
Longitudinal strain
FIGURE 26.7. Fracture plot of the laminate of Figure 26.6.
26.4 ANALYSIS OF STRUCTURES USING THE FINITE ELEMENT METHOD 26.4.1 Introduction
The optimal constitution of a laminate in a given structure cannot be investigated by considering various laminates subjected to a given state of the in-plane resultants (Nx, Ny, Nxy) and bending and twisting moments (Mx, My, Mxy). In practice, the structure to be designed is subjected to given loading and boundary conditions. These conditions impose a field of resultants and moments inside the structure which depends upon the type of laminate which constitutes the structure. To illustrate this fact, we consider the plate in Figure 26.8, clamped along the edges AB and CD, and subjected to distributed loads with resultants of 4 kN and 10 kN applied respectively to the centre of the plate and at points F and G on both sides of the centre. Three types of materials (Figure 26.9) are considered: a material M1 constituted of layers with mat and cloth reinforcement; a material M2 constituted of mat reinforcement layers and unidirectional layers oriented at 0 and 90; a material M3 constituted of mat reinforcement layers and cloth reinforcement layers oriented at 45.
588
z B y
2m H E A G 3m
x F C
D
FIGURE 26.8. Clamped plate and loading.
The characteristics of the layers are: mat reinforcement layers:

EL = ET = 7.67 GPa,
LT = 0.33,
GLT = 2.88 GPa, GLT = 4 GPa,
unidirectional layers:
LT = 0.32,
cloth reinforcement layers:

EL = ET = 16 GPa,
LT = 0.14,
GLT = 2 GPa.
5 4 3 2 1
mat cloth mat cloth mat
1 mm 1.4 mm 2 mm 1.4 mm 1 mm
5 4 3 2 1
0 90 mat 90 0
1 mm 1 mm 2 mm 1 mm 1 mm
3 2 1
45 cloth 2.1 mm mat

2 mm
45 cloth 2.1 mm
laminate M1
laminate M2
laminate M3
FIGURE 26.9. Materials studied in the case of the plate of Figure 26.8.
26.4 Analysis of Structures using the Finite Element Method 1000
589
4000
Bending moment Mx ( N m )
2000
Bending moment Mx ( N m )
5000
500
Coordinate x ( m )
1000
Coordinate x ( m )
laminate M1 laminate M2 laminate M3
Twisting moment Mx ( N m )
5000
500
Coordinate x ( m ) FIGURE 26.10. Variation of moments according to the type of material.
The variations of the moments Mx, My and Mxy along HI (Figure 26.8) have been evaluated using a finite element analysis (following subsection) and are reported in Figure 26.10 for the laminates under consideration. The results obtained clearly show the influence of the nature of the laminate on the variation of the moments. So, these results underline that it is necessary to implement the optimisation of the materials by carrying out a complete analysis of the distribution of strains and stresses inside the structure considered.
26.4.2 Finite Element Method

The analysis of the behaviour of a structure by the finite element method consists in considering the structure as a mesh of elements (Figure 26.11) and
590
element
node
FIGURE 26.11. Meshing of a structure.
establishing the force-displacement relations at the nodes of the elements, taking into account external loading and the boundary conditions imposed to the structure. We then obtain a large dimensional system of linear equations the numerical solution of which leads to the value of the displacement at each node. The stress field is next derived from the displacement field. Finite element analysis requires the preliminary representation of the structure as a mesh of elements. This function is obtained by a preprocessor call structure meshing, which allows the operator to perform an automated meshing of the structure. The meshing can be performed directly in the case of simple shape structures. In the case of complex shape structures, the meshing is performed after geometric modelling based on a formalism of Bezier or Spline type. The analysis of a structure by the finite element method thus appears as one of the steps integrated into a computer aided design system, starting from the definition of the structure (the geometric modelling) to result in designing the structure by the finite element method. The design process is then shown in Figure 26.12. At Section 26.1, it was noted the complexity of the process of optimisation for designing the structures constituted of composite materials, since the optimisation includes the optimisation of the material at the level of the layers and at the level of the layer arrangement. The structure design examples considered hereafter have been carried out with the finite element program PERMAS1. This program is integrated into a computer aided design package including modules for geometric modelling, meshing, finite element analysis and result processing. For composite materials, the finite element program uses finite elements based on the laminate theory that includes the transverse shear effect (Chapter 17) and finite elements based on the theory of sandwich plates (Chapter 18). __________
1. PERMAS Developed by INTES GmbH, Stuttgart, Germany.
26.4 Analysis of Structures using the Finite Element Method
591
Modelling
Meshing
Modification of parameters
Finite element analysis
Comparison with the performances to be obtained
Result analysis
FIGURE 26.12. The design process.
26.4.3 Validation
The use of a finite element program requires performing an estimate of the validity of the numerical results derived from the analysis of the mechanical behaviour of the structure made of composite material. This validity has to be verified by comparing the results obtained by the element finite program in the case of different structures with the results deduced from experimental tests. In the case of simple shape structures, the finite element results can also be compared with the results derived from an analytical investigation. As an example, we have investigated the behaviour of the rectangular plate considered in Subsection 21.2.2.2 using a finite element analysis. The plate has length a = 2.8 m and width b = 0.7 m, is simply supported along the four edges, and is subjected to a uniform pressure of 500 Pa (Figure 21.2). The plate has been divided into 28 elements along its length and 7 elements across its width, which results in 196 elements in total. The finite element investigation leads to a maximum deflection at the centre of the plate equal to: w0max = 5.642 mm whereas the analytical investigation (Relation (21.43)) gives a value of 5.728 mm. The values for the stresses xx and yy obtained by the finite element analysis are compared in Table 26.1 with the values derived by the analytical investigation (Relations (21.45) to (21.50) and Figure 21.3). We observe a very good agreement between the values deduced from finite elements and those obtained by the analytical process (a difference of the order of 1.5 %).
592
TABLE 26.1. Values of the stresses in the layers. z (mm) 1 2.4 3.4 1 2.4
xx (MPa)
Analytical value 0.327 0.785 1.112 0.199 0.477 Finite elements 0.323 0.777 1.099 0.195 0.470
yy (MPa)
Analytical value 0.974 2.339 3.313 1.573 3.774 Finite elements 0.967 2.31 3.28 1.55 3.73
Mat layers
Cloth layers
26.5 EXAMPLES OF PREDESIGNING 26.5.1 Predesigning the Hull of a Yacht

The first example concerns the design of the hull of a yacht (Figure 26.13) 17 m long. The object of the design was to define the choice and the thickness of the materials, so as to have the minimum deformation of the hull without damaging when the mast is shrouded, a deformation evaluated at the fracture limit of the shrouds.
26.5.1.2 Materials
Two materials were to be considered, made of laminated face sheets and a core either of rigid expanded polyvinyl chloride PVC foam (Airex foam) or polypropylene honeycomb (Nidaplast core). The object of the design was to choose the core material and its more suitable thickness, a choice also related to cost and manufacturing conditions, and to determine the final composition of the laminated skins based on layers with glass reinforcements: woven cloth, cross-ply, unidirectional, or mat. At the first step in the design process, three materials were initially considered. These materials (Figure 26.14) have identical skins made of two layers 0.8 mm thick with balanced cloth reinforcement of weight 736 g/m2 and one layer of the same thickness of 0.8 mm with mat reinforcement of weight 400 g/m2. They differ in the core characteristics: a material denoted sandwich 1, with an Airex core 20 mm thick, a material denoted sandwich 2, with a Nidaplast core 20 mm thick, a material denoted sandwich 3, with a Nidaplast core 40 mm thick.
26.5 Examples of Predesigning
593
FIGURE 26.13. Yacht with a sandwich hull.
594
mat cloth cloth Airex or Nidaplast cloth cloth mat
0.8 mm 0.8 mm 0.8 mm
0.8 mm 0.8 mm 0.8 mm
FIGURE 26.14. Sandwich materials considered for the initial design of the yacht hull.
26.5.1.3 Determination of the Mechanical Characteristics of the Materials

The characteristics of the Airex foam and Nidaplast honeycomb have been determined in tensile, compressive and shear tests. The values obtained are: Airex foam: Ec = 70 MPa, Nidaplast honeycomb:
Gc = 25 MPa,
c =
Ec 1 = 0.4, 2Gc
(26.38)
(26.39)
The elasticity moduli of the skins have been deduced from tensile tests carried out on the sandwich materials oriented at 0, 45 and 90 to the material directions of the skins (Section 11.4):
ELs = ETs = 13.9 GPa, GLTs = 2.2 GPa,
LTs = 0.16,
(26.40)
An analytical evaluation (Chapter 15, Section 15.2) leads to the following values of the moduli: ELs = ETs = 14.2 GPa, GLTs = 2.5 GPa, LTs = 0.15, (26.41) values that are in good agreement with the experimental values.
26.5.1.4 Validation of the Model of the Mechanical Behaviour of the Sandwich Materials
The validation was carried out in the case of three-point bending tests and in the case of a plate supported at three points and subjected to a load at its centre.
595
1. Three-point bending tests
The results established in Section 20.4, in the framework of the sandwich theory, show that in a three-point bending test the relation between the deflection wc at the centre and the applied load is expressed as a function of the span length L between the supports by Relation (20.103):
w c PL3 12 = D11 + 2 F55 . P 48b L
(26.42)
This relation can be written in the form (20.162), which leads to:
wc A 2 B = L + , PL ELs Gc
(26.43)
where the parameters A and B are expressed as functions of b, h and h1 in accordance with Relation (20.162). Figure 26.15 shows the experimental results obtained in the case of sandwiches with a Nidaplast core. The experimental values aligned along straight lines confirm the validity of Relation (26.43) and lead to: sandwich 2 (Nidaplast, h = 20 mm):
ELs = 13.3 GPa, Gc = 12.8 MPa, Ga = 9.4 MPa.
(26.44)
sandwich 3 (Nidaplast, h = 40 mm):

ELs = 13.8 GPa,
(26.45)
Deflection wc / PL ( 10 N )
7 6 5 4 3 2 1 1 0 0.1 0.2
h = 20 mm
h = 40 mm
0.3
0.4
0.5
0.6
2
0.7
0.8
0.9
Square of the span length L
(m )
FIGURE 26.15. Deflection at the beam centre as a function of the span length between supports in a three-point bending test (sandwich materials with Nidaplast core).
596
A similar process applied to sandwich 1 (Airex foam, h = 20 mm) leads to:

ELp = 13,3 GPa, Ga = 28 MPa. (26.46)
The set of the values deduced from the three-point bending tests carried out in the case of the sandwich materials, (26.44) to (26.46), is consistent with the values (26.38) and (26.39) evaluated initially in the case of the sandwich constituents.
2. Three-point supported plate
The validation of the mechanical behaviour has been also carried out in the case of a plate supported at points A, B, C, and subjected to a load FD applied at D (Figure 26.16). The deflection wD, measured at D, is compared with the value calculated by the finite element analysis with the values of the moduli determined previously. The values obtained for the deflection wD in the case of a load FD = 2000 N are: Measurements (mm) Sandwich 1 Sandwich 2 Sandwich 3 5.76 9.32 5.20 Finite elements (mm) 5.54 8.65 4.92
These results show deviations of 4 to 8 % between the values measured and the values derived from the finite element analysis. These differences allow us to assess the validity of modelling the mechanical behaviour of the materials considered using the sandwich theory associated with the finite element analysis.
26.5.1.5 Predesigning
The first analyses implemented by finite element analysis on the yacht hull show the necessity of reinforcing the hull with sandwich beams and stringers of 1m B FD C
0.8 m
FIGURE 26.16. Three-point supported plate subjected to a concentrated load.
597
FIGURE 26.17. Deformed shape of the yacht hull when the mast is shrouded.
hat section in the vicinity of the support of the mast. Figures 26.17, 26.18 and 26.19 give examples obtained in the course of the predesigning: the deformed shape of the hull (Figure 26.17) and the contour plot of the stresses, when the mast is shrouded. The successive analyses, associated with considerations of manufacturing cost, led finally to manufacturing the hull with Airex core sandwich and the deck with polypropylene honeycomb core sandwich.
26.5.2 Predesigning the Hood of a Car

The second example concerns the design of the hood of a car, conducted by an approach similar to the preceding one.
26.5.2.1 Material
The material used is a sandwich (Figure 26.20) constructed from a core of polyurethane foam and two skins of glass mat polyurethane foam. This sandwich material was manufactured by expansion of the foam simultaneously in the core and in the skins. The object of the design was to optimise the material thickness
598
FIGURE 26.18. Stress distribution in the hull when the mast is shrouded.
FIGURE 26.19. Stress distribution inside the structure of the hull when the mast is shrouded.
599
glass mat polyurethane
polyurethane foam
FIGURE 26.20. Sandwich used for the car hood.
(thickness of the core, thickness of the skins), the volume fraction of the fibres and the density of the polyurethane foam. The characteristics of the material initially studied were: thickness of the sandwich: ht = 16 mm, thickness of the skins: h1 = 3 mm, density of the polyurethane: 100 kg/m3, in each skins, two glass mats of weight: Ms = 450 g/m2.
26.5.2.2 Determination of the Mechanical Characteristics of the Sandwich Material

The mechanical characteristics of the polyurethane foam were measured in tensile, compressive and shear tests. The values obtained are: elasticity moduli:
Ec = 78 MPa, Gc = 27 MPa,
c = 0.45,
(26.47)
fracture stresses:
tensile strength
tc = 1.5 MPa , compressive strength sc = 0.8 MPa ,

Sc = 1 MPa .
(26.48)
shear strength
The elasticity moduli of the skins were deduced from tensile, compressive and shear tests carried out on the sandwich material. The values obtained are:
ELm = ETm = 2,800 MPa, GLTm = 1,200 MPa,
LTm = 0.35. (26.49)
An analytical evaluation as a function of the characteristics of the constituents (Chapter 15, Subsection 15.2.6) leads to the values:
ELm = ETm = 2,888 MPa, GLTm = 1,080 MPa,
LTm = 0.33, (26.50)
values in good agreement with the experimental ones.
600
26.5.2.3 Modelling the Mechanical Behaviour of the Sandwich Material

As previously, the study of the mechanical behaviour of the sandwich material was conducted in the case of three-point bending tests and in the case of a plate of large size subjected to a mechanical loading.
1. Three-point bending tests
In the case of three-point bending tests, the experimental results as a function of the span length L between the supports are reported in Figure 26.20. The application of Relation (26.43) leads to:
ELm = 4,500 MPa, Gc = 46 MPa.
(26.51)
These values do not agree with the values (26.47), (26.49) and (26.50). Thus there appears here a difficulty which arises from the fact of the thickness of the skins being not very much smaller than the thickness of the core, which admits Relation (26.43). Modelling the mechanical behaviour of the mat-polyurethane material by the sandwich plate theory leads to the value of a fictitious modulus (4,500 MPa) much higher than the actual modulus (2,800/2,900 MPa), the shear modulus of the core also being higher. The analytical investigation with the laminate theory which includes the effect of the transverse shear deformation leads to Relation (20.163) between the deflection wc at the centre and the applied load P. In the case of the material under consideration, this relation can be written in the form: A wc L2 + = PL D ELm B , h1 Gm Gc 1 + 2 h Gc
D = 1+
4h h1 h + 3 1 , h h + h1
(26.52)
Deflection wc / PL ( 10 N )
1 5
7 6 5 4 3 2 1 0 0.02 0.04
2
0.06
0.08
2
Square of the span length L
(m )
FIGURE 26.21. Deflection at the beam centre as a function of the span length in threepoint bending tests in the case of mat-polyurethane sandwich.
601
where Gm is the transverse shear modulus of the skins. Expressions (26.43) and (26.52) coincide in the case of small thicknesses of the skins (h1 h) . Applying Relation (26.52) to the experimental results of Figure 26.21, leads to: ELm = 3,400 MPa, Gc = 33 MPa. (26.53) The Young modulus ELm is reduced to 3,400 MPa, but still remains higher than the actual modulus 2,800/2,900 MPa. Lastly, a finite element analysis using volume elements with the actual moduli leads to results which are practically the same as the experimental results. Thus, it appears that the mechanical behaviour of the sandwich material can be described in three-point bending tests by the sandwich or laminate theory with transverse shear, but on introducing a fictitious modulus for the skins (ELm = 4,500 MPa for the sandwich model, ELm = 3,400 MPa for the laminate theory), which differs from the actual model. Modelling of the mechanical behaviour with the actual values of the moduli requires the use either of a laminate theory of order higher than 1 [34, 35] or a finite element analysis with volume elements, at the cost of a greater complexity of these analyses.
2. Plate bending
The mechanical behaviour of the sandwich material was also been studied in the case of plate clamped along one edge and subjected to a concentrated load applied at one of the points A, B, . . . , H (Figure 26.22). The deflection measured at the different points (A, B, . . . , H) was compared with the values deduced from a finite element analysis with the moduli determined previously in the modelling considered. Tables 26.2 and 26.3 compare some of the values obtained. For comparison, there are also reported the values obtained by considering for each model the actual moduli of the skins. As in the case of three-point bending, the mechanical behaviour of the plate can be described by the sandwich theory or the laminate theory with transverse shear by using the fictitious values of the modulus
0.62 m
E A B
F G C D
H
0.150 m 0.510 m 0.255 m
FIGURE 26.22. Clamped plate loaded at different points.
602
TABLE 26.2. Load of 10 N at point C of the plate in Figure 26.22. Model ELm (MPa) Deflection (mm) at: C A D Sandwich 4,500 2,900 Laminate 3,400 2,900 Volume elements 2,900 Experimental values
0.575 0.089 0.822
0.890 0.153 1.291
0.590 0.092 0.856
0.691 0.108 1.012
0.580 0.096 0.841
0.595 0.092 0.875
TABLE 26.3. Load of 10 N at point G of the plate in Figure 26.22. Model ELm (MPa) Deflection (mm) at: Sandwich 4,500 2,900 Laminate 3,400 2,900 Volume elemnts 2,900 Experimental values
A D E G H
0.097 0.801 0.116 0.648 0.908
0.153 1.258 0.184 1.024 1.429
0.092 0.834 0.142 0.656 0.924
0.108 0.976 0.166 0.768 1.081
0.094 0.812 0.122 0.623 0.882
0.099 0.841 0.135 0.639 0.89
ELm given respectively in (26.51) and (26.54). The results derived from the finite element analysis using volume elements and the actual value (26.50) of the modulus of the skins agree with the experimental values. The introduction of this value in the analysis with sandwich elements or laminate elements leads to some notable differences, but nevertheless less with the laminate theory including transverse shear (a difference of about 17 %) than with sandwich theory (a difference of about 50 %).
26.5.2.4 First Steps of the Design

Among the imposed mechanical specifications, predesigning of the hood was carried out by considering the following characteristics: deformation without damage to the closed hood subjected to a concentrated load ; low deformation of the hood under its own weight; deformation without any damage of the open hood subjected to a distributed load ( the action of wind upon the open hood); resistance to a lateral impact; etc.
603
Figure 26.23 shows the geometric modelling and the meshing used to analyse the mechanical behaviour of the hood. The hood opens forward by means of two links at A and B. When closed, the hood is supported at points C, D and E. The maximum deflection evaluated by finite element analysis for the deformation of the hood under its own weight is some 4 mm. Figure 26.24 shows the deformation of the hood under a lateral impact. Figures 26.25 and 26.26 show the contour plots of the stresses obtained according to Von Mises failure criterion (Figure 26.25) and the maximum tensile stress criterion (Figure 26.26), in the case of the hood closed and subjected to a concentrated load of 1 kN. The values of the stresses obtained lead to a safety factor of the order of 3 for a first damaging of the sandwich material. The following stages of the mechanical analysis led to an optimised design of the hood with ribs and different thicknesses for the sandwich material according to the load distribution across the hood.
26.5.3 Conclusions on Predesigning

The two examples considered previously underline the necessity for the designers to implement mechanical tests for characterising the materials under consideration and to validate the mechanical behaviour of the composite structure derived from the finite element analysis. The tests on the materials are carried out on the constituents: layers in the case of laminate materials, skins and core in the case of sandwich materials. In the case of predesigning, the characteristics of laminate or sandwich materials can also be
D A
FIGURE 26.23. Geometric modelling and meshing of the hood.
604
FIGURE 26.24. Deformed shape of the hood when subjected to a lateral impact.
evaluated by considering the analytical approaches developed in Parts 3 and 4. These approaches allow then to estimate rapidly the influence of the different parameters as the mechanical characteristics of the constituents, their proportions, the structure of the material as a function of the layer stacking sequence, etc. The global and local mechanical behaviour of the composite structure is next evaluated using a finite element analysis. In the examples investigated previously, we focused the attention of the designer on the necessity to validate the finite element analyses by comparing the obtained results with experimental results derived from simple shape structures. Once this validation is carried out, the finite element of the composite structure will be implemented using modelling the best fitted to the materials under consideration.
605
FIGURE 26.25. Stress distribution in the hood subjected to a concentrated load, according to Von Mises failure criterion.
FIGURE 26.26. Stress distribution in the hood subjected to a concentrated load, according to the maximum tensile stress criterion.
APPENDIX A
Polynomial Function of a Beam with Clamped Ends
The polynomial function, which satisfy the clamping conditions at the two ends u = 0 and u = 1 of a beam, is expressed as:
X m (u ) = u 2 ( u 1) u m 1 ,
(A.1)
where u is the reduced variable

u=
and a is the length of the beam.
x , a
(A.2)
0 TABLE A.1. Values of the integrals I m =
1 0
Xm du .
7 0.202 020 8 0.151 515
m
0 Im
1 3.333 333
2 1.666 667
3 0.952 381
4 0.595 238
5 0.396 825
6 0.277 778
0 ( Im = values in the table 102 ).
00 TABLE A.2. Values of the integrals I mi =
1 0
Xm Xi d u .
6 7 0.066 600 0.045 788 0.032 321 0.023 343 0.017 200 0.012 900 0.009 828 0.007 595 8 0.045 788 0.032 321 0.023 343 0.017 200 0.012 900 0.009 828 0.007 595 0.005 944
m/i 1 2 3 4 5 6 7 8
1 1.587 302 0.793 651 0.432 900 0.252 525 0.155 400 0.099 900 0.066 600 0.045 788
2 0.793 651 0.432 900 0.252 252 0.155 400 0.099 900 0.066 600 0.045 788 0.032 321
3 0.432 900 0.252 252 0.155 400 0.099 900 0.066 600 0.045 788 0.032 321 0.023 343
4 0.252 525 0.155 400 0.099 900 0.066 600 0.045 788 0.032 321 0.023 343 0.017 200
5 0.155 400 0.099 900 0.066 600 0.045 788 0.032 321 0.023 343 0.017 200 0.012 900
0.099 900 0.066 600 0.045 788 0.032 321 0.023 343 0.017 200 0.012 900 0.009 828
00 3 ( I mi = values in the table 10 ).
608
Appendix A Polynomial Function of a Beam with Clamped Ends
TABLE A.3. Values of the integrals

m/i
1 2 3 4 5 6 7 8 1 2 3 4
02 I mi =
1 0
Xm
d2 X i d u2
6
20 02 d u. ( I mi = I mi ).
1.904762 0.952381 0.476190 0.238095 0.115440 0.050505 0.015540 0.003330 0.952381 0.634921 0.396825 0.245310 0.151515 0.093240 0.056610 0.033300 0.476190 0.396825 0.288600 0.202020 0.139860 0.096570 0.066600 0.045788 0.238095 0.245310 0.202020 0.155400 0.116550 0.086580 0.064103 0.047404 0.115440 0.151515 0.139860 0.116550 0.093240 0.073260 0.057100 0.044351 0.050505 0.093240 0.096570 0.086580 0.073260 0.060331 0.049020 0.039560 0.015540 0.056610 0.066600 0.064103 0.057100 0.049020 0.041280 0.034400 0.003330 0.033300 0.045788 0.047404 0.044351 0.039560 0.034400 0.029485

1 d2 X 0
22 TABLE A.4. Values of the integrals I mi =
du
m 2
d2 X i d u2
d u.
7 8 0.036364 0.067133 0.086314 0.095904 0.098901 0.097738 0.094118 0.089164
m/i
1 2 3 4 5 6 7 8
1 0.800000 0.400000 0.228571 0.142857 0.095238 0.066667 0.048485 0.036364
2 0.400000 0.342857 0.257143 0.190476 0.142857 0.109091 0.084848 0.067133
3 0.288571 0.257143 0.228571 0.190476 0.155844 0.127273 0.104429 0.086314
4 0.142857 0.190476 0.190476 0.173160 0.151515 0.130536 0.111888 0.095904
5 0.095238 0.142857 0.155844 0.151515 0.139860 0.125874 0.111888 0.098901
6 0.066667 0.109091 0.127273 0.130536 0.125874 0.117483 0.107692 0.097738
0.048485 0.084848 0.104429 0.111888 0.111888 0.107692 0.101357 0.094118
01 TABLEAU A.5. Values of the integrals I mi =
1 0
Xm
d Xi d u. du
7 2,997003 1,665002 0,915751 0,484809 0,233427 0,085999 0 8 2,331002 1,373626 0,808016 0,466853 0,257998 0,128999 0,049142 0
m/i
1 2 3 4 5 6 7 8
1 0 7,936508
2 7,936508 0
3 7,936508 2,164502 0
4 6,493506 2,525525 0,777001 0
5 5,050505 2,331002 0,999001 0,333000 0
6 3,885004 1,998002 0,999001 0,457875 0,161603 0
7,936508 2,164502
6,493506 2,525525 0,777001
5,050505 2,331002 0,999001 0,333000
3,885004 1,998002 0,999001 0,457875 0,161603
2,997003 1,665002 0,915751 0,484809 0,233427 0,085999
2,331002 1,373626 0,808016 0,466853 0,257998 0,128999 0,049142
609
12 = TABLE A.6. Values of the integrals I mi
1dX 0
d2 X i
d u d u2
6
d u.
7 8
0.979021 0.809191 0.599401 0.412088 0.261797 0.147059 0.061920 0
m/i
1 2 3 4 5 6 7 8
1
0 2.857143
2
2.857143 0
3
2.857143 0.952381 0
4
2.380952 1.190476 0.432900 0
5
1.904762 1.168831 0.606061 0.233100 0
1.515151 1.060606 0.652681 0.349650 0.139860 0
1.212121 0.932401 0.639361 0.399600 0.219780 0.090498 0
2.857143 0.952381
2.380952 1.190476 0.432900
1.904762 1.168831 0.606061 0.233100
1.515151 1.060606 0.652681 0.349650 0.139860
1.212121 0.932401 0.639361 0.399600 0.219780 0.090498
0.979021 0.809191 0.599401 0.412088 0.261797 0.147059 0.061920
2 ( I 12 mi = values in the table 10 ).
APPENDIX B
Characteristic Function of a Beam with Clamped Ends
The characteristic function of the transverse vibrations of a beam, satisfying to the clamping conditions at the two ends of the beam is expressed as: X m (u ) = cos mu cosh mu m ( sin mu sinh mu ) , wher u is the reduced
u= x , a
(B.1)
(B.2)
and a is the length of the beam. The values of the parameters m and m for m increasing from 1 to 8 are reported in Table 21.3.
22 TABLE B.1. Values of the integrals I mi =
1 d2 X 0
du
m 2
d2 X i d u2
d u.
i=m
22 I mi
1 2 3 4 5 6 7 8
500.564 3 803.537 14 617.630 39 943.799 89 135.407 173 881.316 308 208.452 508 481.543
22 I mi = 0 if i m.
612
Appendix B Characteristic Function of a Beam with Clamped Ends
1 0
Xm
6 0
d2 X i d u2
d u.
7 6.10804 0 22.9842 0 38.0302 0 508.041
0.00025
m/i
1 2 3 4 5 6 7 8
1 12.3026 0 9.73079 0 7.61544 0 6.10804 0
2 0 46.0501 0 17.1289 0 15.1946 0 13.1366
3 9.73079 0 98.9048 0 24.3490 0 22.9842 0.00011
4 0 17.1289 0 171.586 31.2764 0 30.5784
5 7.61544 0 24.3499 0 263.998 0 38.0302 0.00019
8 0 13.1366 0.00011 30.5784 0.00019 44.6689 0.00025 695.672
15.1946 0 31.2764 0 376.150 0 44.6689
20 20 I mi = I mi .
11 I mi =
1 dX 0
dXi 02 d u = I mi . du du
m
0 TABLE B.3. Values of the integrals I m =
1 0
Xm du .
m
1 2 3 4 5 6 7 8
0 Im
0.8308615 0 0.3637694 0 0.2314981 0 0.1697653 0.0000011
613
1 0
Xm
d Xi d u. du
7 0 899817 0 3.308823 0 13.787164 0 15.813463 8 0.251251 0 1.396553 0 4.061433 0 15.813464 0
m/i
1 2 3 4 5 6 7 8
1 0 3.342016 0 0.906926 0 0.430472 0 0.251251
2 3.342016 0 5.516101 0 1.726226 0 0.899817 0
3 0 5.516101 0 7.632796 0 2.532434 0 1.396553
4 0.906926 0 7.632796 0 9.703955 0 3.308823 0
5 0 1.726226 0 9.703955 0 11.752299 0 4.061433
6 0.430472 0 2.532434 0 11.752299 0 13.787164 0
10 01 I mi = I mi .
1dX 0
d2 X i
d u d u2
6 40.8514 0
d u.
7 0 8 31.1808 0 410.0485 0 1873.970 0 9949.642 0
m/i
1 2 3 4 5 6 7 8
1 0 122.0650 0 559.5845 0 40.8514 0 31.1808
2 122.0650 0 476.6737 0 234.4200 0 166.6288 0
3 0 476.6737 0 1186.455 0 568.5990 0 410.0485
4 559.5845 0 1186.455 0 2370.414 0 1102.168 0
5 0 234.4200 0 2370.414 0 4146.662 0 1873.970
166.6288 0 1102.168 0 6633.601 0 9949.642
568.5990 0 4146.662 0 6633.601 0
21 I mi = I 12 mi .
References
[1] J.-M. Berthelot (2006). Matriaux composites. Comportement mcanique et analyse des structures. 4th edition. Tech & Doc, Paris. [2] J.-M. Berthelot (1999). Composite Materials. Mechanical Behavior and Structural Analysis. Springer, New York. [3] J.-M. Berthelot (2007). Dynamics of Composite Materials and Structures. Le Mans. Available online at http:/www.CompoMechAsia.org. [4] C.C. Chamis et G.P. Sendeckyj (1968). Critique on theories predicting thermoelastic properties of fibrous composites. J. Compos. Mat. (July), 332358. [5] R.M. Christensen (1979). Mechanics of Composite Materials. John Wiley & Sons, New York. [6] Z. Hashin (1965). On elastic behaviour of fibre reinforced materials of arbitrary transverse plane geometry. J. Mech. Phys. Solids, 13, 119. [7] R. Hill (1964). Theory of mechanical properties of fibre-strengthened materials : I. Elastic behavior. J. Mech. Phys. Solids, 12, 199. [8] Z. Hashin et B.W. Rosen (1964). The elastic moduli of fiber-reinforced materials. J. Appl. Mech. (June), 223-232. [9] Z. Hashin (1966). Viscoelastic fiber reinforced materials. AIAA J., 4, 1411. [10] J.J. Hermans (1967). The elastic properties of fiber reinforced materials when the fibers are aligned. Proc. K. Ned. Akad. Wet., B70, 1. [11] R.M. Christensen et K.H. Lo (1979). Solutions for effective shear properties in three phase sphere and cylinder models. J. Mech. Phys. Solids, 27(4)4. [12] J.C. Halpin et S.W. Tsai (1969). Effects of environmental factors on composite materials. AFML-TR 67-243, (June). [13] D.F. Adams et D.R. Doner (1967). Transverse normal loading of a unidirectional composite. J. of Compo. Mat. (April), 152-164. [14] D.F. Adams et D.R. Doner (1967). Longitudinal shear loading of a unidirectional composite. J. Compos. Mat. (January), 4-17.
616
References
[15] R. Hill (1950). The Mathematical Theory of Plasticity. Oxford University Press, London. [16] V.D. Azzi et S.W. Tsai (1965). Anisotropic Strength of Components. Exper. Mech., 5, 286-288. [17] O. Hoffman (1967). The Brittle Strength of Orthotropic Materials. J. Compos. Mat., 1, 200-206. [18] S.W. Tsai et E.M. Wu (1971). A General Theory of Strength for Anisotropic Materials. J. Compos. Mat., 5, 58-80. [19] J.C. Halpin, K. Jerine et J.M. Whitney (1971). The Laminate Analogy for 2 and 3 Dimensional Composite Materials. J. Compos. Mat., 5, 36-49. [20] E. Reissner (1945). The effect of transverse shear deformation on the bending of elastic plates. J. Appl. Mech., 12, 69-77. [21] R.D. Mindlin (1951). Influence of rotatory inertia and shear on flexural motions of isotropic, elastic plates. J. Appl. Mech., 18, 336-343. [22] J.M. Whitney et N.J. Pagano (1970). Shear deformation in heterogeneous anisotropic plates. J. Appl. Mech., 37, 1031-1036. [23] J.M. Whitney (1972). Stress analysis of thick laminated composite and sandwich plates. J. Compos. Mat., 6, 426-440. [24] N.J. Pagano (1969). Exact Solutions for Composite Laminates in Cylindrical Bending. J. Compos. Mat., 3, 398-411. [25] S.P. Timoshenko (1955). Strength of Materials, Vol. 1, 3me dition. Von Nostrand , Princeton. [26] D. Young (1950). Vibration of Rectangular Plates by Ritz Method. J. Appl. Mech., 17, 448-453. [27] J.M. Whitney (1987). Structural Analysis of Laminated Anisotropic Plates. Technomic Publishing Company, Lancaster. [28] S. Timoshenko, D.H. Young et W. Weaver, Jr (1974). Vibration Problems in Engineering, 4me dition. John Wiley & Sons, New York, Londres, Sydney, Toronto. [29] J.E. Ashton et M.E. Waddoups (1969). Analysis of Anisotropic Plates. J. Compos. Mat., 3, 148-165. [30] S. Timoshenko et J.M. Gere (1961). Theory of Elastic Stability. Mc GrawHill, New York. [31] J. Crank (1975). The Mathematics of Diffusion, 2nd edition. Oxford University Press, Oxford. [32] P. Cirese, M. Marchetti et S. Sgubini (1990). Design and manufacturing criteria for high precision composite antenna reflectors. Prediction of the residual distortions after the manufacturing process. Compos. Struct., 16, 209-235.
References
617
[33] J.-M. Berthelot (1995). High mechanical performance composites and design of composite structures. Polymers and Other Advanced Materials, ds. P.N. Prasad et al., Plenum Press, New York. [34] K.H. Lo, R.M. Christensen et E.M. Wu (1977). A High-Order Theory of Plate Deformation. Part 2 : Laminated Plates. J. Appl. Mech., 44, Trans. ASME, Ser. E, 99, 669-676. [35] B.N. Pandya et T. Kant (1987). Finite Element Analysis of Laminated Composite Plates using a High-Order Displacement Model. Compos. Sci. Technol., 32, 137-155.
Jean-Marie Berthelot
Mechanical Behaviour of Composite Materials and Structures
Mechanical Behaviour of Composite Materials and Structures is a textbook which provides the fundamental elements which are necessary for understanding the mechanical behaviour of Composite Materials and Composite Structures. In this treatise, the different parts have been developed in such a way to have a continuity in the concepts and theories, providing thus a unified fundamental approach. After the presentation of the constitution of composite materials, the textbook develops progressively the different concepts needed for modelling the mechanical behaviour of structures constructed with laminate or sandwich composite materials. Next, the problems of the design of composite structures are considered showing how the concepts can be applied. Content and progression of the book are developed considering four principal objectives: to consider the composite material as a usual material; to integrate the contribution of the computer aided design for solving the problems of design of composite structures; to introduce the different concepts in a progressive way in such a way to make easy the understanding of the concepts for the reader; to confront the results derived from modelling with the actual behaviour of the materials and structures. Exercises are proposed at the end of the chapters, to illustrate the application of the concepts introduced. Mechanical Behaviour of Composite Materials and Structures constitutes a reference book for students, engineers and research workers.
Jean-Marie Berthelot is an Emeritus Professor at the Institute for Advanced Materials and Mechanics (ISMANS), Le Mans, France. His current research is on the mechanical behaviour of composite materials and structures. He has published extensively in the area of composite materials and is the author of a textbook entitled Composite Materials, Mechanical Behavior and Structural Analysis published by Springer, New York, in 1999.

Berthelot SF 98 99 175 181 PDF

Uploaded by

Copyright:

Available Formats

Berthelot SF 98 99 175 181 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Berthelot SF 98 99 175 181 PDF

Uploaded by

Copyright:

Available Formats

Part I

Constituents and Architecture of Composite Materials

Basic Elements on Composite Materials

1.1 COMPOSITE MATERIALS 1.1.1 Definition

1.1.2 General Features

Chapter 1 Basic Elements on Composite Materials

TABLE 1.1. Examples of composite materials, in a general sense.

Cement/sand/granulates Carbon/carbon fibres Ceramic/ceramic fibres

Aluminium/boron fibres Aluminium/carbon

FIGURE 1.1. Composite Material.

1.2 Classification of Composite Materials

1.2 CLASSIFICATION OF COMPOSITE MATERIALS

1.2.1 Classification according to the Form of Constituents

1.2.1.1 Fibre Composites

Chapter 1 Basic Elements on Composite Materials

1.2.1.2 Particle Composites

1.2.2 Classification according to the Nature of Constituents

1.3 Why Composite Materials?

1.3 WHY COMPOSITE MATERIALS ?

1.3.1 Specific Mechanical Characteristics

Chapter 1 Basic Elements on Composite Materials

FIGURE 1.2. Tensile loading of a beam.

1.3 Why Composite Materials?

1.3.2 Mechanical Characteristics of Materials

Chapter 1 Basic Elements on Composite Materials

1.3.3 Composite Materials

1.4 VOLUME AND WEIGHT FRACTIONS 1.4.1 Introduction

1.4 Volume and Weight Fraction

TABLE 1.2. Specific properties of usual materials in bulk form.

(MPa) 3402,100 140620 700-2,100 1,1004,100 700

(kN m/kg) 43270 52230 280840 57210 380

Steel Aluminium alloys Wood Glass Tungsten Beryllium

(MPa) 3,500 4,600 2,100 3,500 2,800 2,800

(kN m/kg) 1,380 1,850 1,100 1,890 1,870 1,100

1.4.2 Volume Fractions

Chapter 1 Basic Elements on Composite Materials

1.4.3 Weight Fractions

(1.13) (1.14) (1.15)

1.4.4 Relations between Volume and Weight fractions

(1.16) (1.17) (1.18)

The total weight of composite material is:

1.4 Volume and Weight Fraction

Whence the expression for the specific weight of the composite:

Chapter 1 Basic Elements on Composite Materials

1.4.5 Presence of Porosity

The Constituent Elements of Composite Materials

2.2 THE RESINS 2.2.1 Types of Resins

Chapter 2 The Constituent Elements of Composite Materials

2.2.2 Thermosetting Resins

2.2.2.1 Polyesters Resins

2.2 The Resins

2.2.2.2 Condensation Resins

Chapter 2 The Constituent Elements of Composite Materials

2.2.2.3 Epoxide Resins

2.2 The Resins

2.2.3 Thermoplastic Resins

Chapter 2 The Constituent Elements of Composite Materials

2.2.4 Thermostable Resins

2.3 FILLERS AND ADDITIVES 2.3.1 Introduction